Drift Localization.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "# wget --no-check-certificate -r 'https://docs.google.com/uc?export=download&id=14UBHX6GTW_4YiyjNJB3EEq7Xb83AjuaK' -O process_mining_datasets.zip\n",
    "\n",
    "import pandas as pd\n",
    "# pd.set_option(\"max_columns\", 200)\n",
    "import numpy as np\n",
    "from tqdm import tqdm\n",
    "import json\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "%load_ext autoreload\n",
    "%autoreload 2\n",
    "\n",
    "# Meus pacotes\n",
    "import parse_mxml as pm\n",
    "import log_representation as lr\n",
    "import plots as plts\n",
    "import model_utils as mu\n",
    "import drift_detection as dd\n",
    "import offline_streaming_clustering as off_sc\n",
    "from sklearn.model_selection import ParameterGrid\n",
    "from sklearn.metrics import roc_auc_score, ndcg_score, dcg_score, roc_curve, average_precision_score\n",
    "\n",
    "from scipy.spatial import distance\n",
    "from sklearn.base import clone as sk_clone \n",
    "from scipy.optimize import linear_sum_assignment\n",
    "import traceback \n",
    "\n",
    "from copy import deepcopy\n",
    "import random\n",
    "random.seed(42)\n",
    "import os\n",
    "import re\n",
    "import warnings\n",
    "warnings.filterwarnings(\"ignore\")\n",
    "from scipy.spatial import distance\n",
    "import glob\n",
    "\n",
    "from sklearn.cluster import KMeans, AgglomerativeClustering\n",
    "\n",
    "import gc\n",
    "gc.enable()\n",
    "\n",
    "def string_to_numpy(text, dtype=None):\n",
    "    \"\"\"\n",
    "    Convert text into 1D or 2D arrays using np.matrix().\n",
    "    The result is returned as an np.ndarray.\n",
    "    https://stackoverflow.com/questions/50589770/ast-literal-eval-malformed-node-or-string-while-converting-a-string-with-list\n",
    "    \"\"\"\n",
    "    import re\n",
    "    text = text.strip()\n",
    "    # Using a regexp, decide whether the array is flat or not.\n",
    "    # The following matches either: \"[1 2 3]\" or \"1 2 3\"\n",
    "    is_flat = bool(re.match(r\"^(\\[[^\\[].+[^\\]]\\]|[^\\[].+[^\\]])$\",\n",
    "                            text, flags=re.S))\n",
    "    # Replace newline characters with semicolons.\n",
    "    text = text.replace(\"]\\n\", \"];\")\n",
    "    # Prepare the result.\n",
    "    result = np.asarray(np.matrix(text, dtype=dtype))\n",
    "    return result.flatten() if is_flat else result\n",
    "\n",
    "def opts_to_string(opts):\n",
    "    return \"_\".join([\n",
    "        opts[\"tipo_mudanca\"] + str(opts[\"log_size\"]/1000).replace(\".0\", \"\") + \"k\",\n",
    "        str(opts[\"log_size\"]),\n",
    "        opts[\"model\"],\n",
    "        opts[\"representation\"],\n",
    "        str(opts[\"window_size\"]),\n",
    "        str(opts[\"sliding_window\"])\n",
    "    ])\n",
    "\n",
    "# # # # # # # # # # #\n",
    "# LOAN APPLICATIONS #\n",
    "# # # # # # # # # # #\n",
    "aliases = {\n",
    "    'Loan__application_received': 'START',\n",
    "    'Appraise_property': 'A',\n",
    "    'Approve_application': 'B',\n",
    "    'Assess_eligibility': 'C',\n",
    "    'Assess_loan_risk': 'D',\n",
    "    'Cancel_application': 'E',\n",
    "    'Check__application__form_completeness': 'F',\n",
    "    'Check_credit_history': 'G',\n",
    "    'Check_if_home_insurance_quote_is_requested': 'H',\n",
    "    'Prepare_acceptance_pack': 'I',\n",
    "    'Receive_updated_application': 'J',\n",
    "    'Reject_application': 'K',\n",
    "    'Return_application_back_to_applicant': 'L',\n",
    "    'Send_acceptance_pack': 'M',\n",
    "    'Send_home_insurance_quote': 'N',\n",
    "    'Verify_repayment_agreement': 'O',\n",
    "    'Loan__application_approved': 'END_A',\n",
    "    'Loan_application_rejected': 'END_R',\n",
    "    'Loan__application_canceled': 'END_C',\n",
    "}\n",
    "\n",
    "inv_aliases = {v: k for k, v in aliases.items()}\n",
    "\n",
    "objects = {\n",
    "    \"representation\": {\n",
    "        \"binary\": lambda x: lr.get_binary_representation(lr.get_traces_as_tokens(x)),\n",
    "        \"frequency\": lambda x: lr.get_frequency_representation(lr.get_traces_as_tokens(x)),\n",
    "        \"tfidf\": lambda x: lr.get_tfidf_representation(lr.get_traces_as_tokens(x)),\n",
    "        \"transitions\": lambda x: lr.get_binary_transitions_representation(lr.get_traces_as_tokens(x)),\n",
    "        \"positions\": lambda x: lr.get_positions_representation(lr.get_traces_as_tokens(x)),\n",
    "        \"positions_min_max\": lambda x: lr.get_min_max_positions_representation(lr.get_traces_as_tokens(x)),\n",
    "        \"positions_avg\": lambda x: lr.get_positions_representation(lr.get_traces_as_tokens(x), include_cols=[\"_avg\"]),\n",
    "    }\n",
    "}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "base_dir = \"C:/Users/rafae/OneDrive/Documentos/resultados_clusterizacoes/\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "def translateLabels(masterList, listToConvert):    \n",
    "    contMatrix = contingency_matrix(masterList, listToConvert)\n",
    "    labelMatcher = munkres.Munkres()\n",
    "    labelTranlater = labelMatcher.compute(contMatrix.max() - contMatrix)\n",
    "\n",
    "    uniqueLabels1 = list(set(masterList))\n",
    "    uniqueLabels2 = list(set(listToConvert))\n",
    "\n",
    "    tranlatorDict = {}\n",
    "    for thisPair in labelTranlater:\n",
    "        tranlatorDict[uniqueLabels2[thisPair[1]]] = uniqueLabels1[thisPair[0]]\n",
    "    \n",
    "    return tranlatorDict\n",
    "    #return [tranlatorDict[label] for label in listToConvert]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.metrics.cluster import contingency_matrix\n",
    "import munkres"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{0: 8, 5: 1, 1: 2, 8: 3}"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "list1 = [1, 1, 0, 5, 5, 1, 8, 1]\n",
    "list2 = [3, 3, 1, 2, 2, 3, 8, 3]\n",
    "\n",
    "translateLabels(list2, list1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "binary_support = [\"cf\", \"cb\", \"cm\", \"rp\", \"pm\", \"fr\"]\n",
    "binary_not_support = [\"pl\", \"cd\", \"cp\", \"sw\"]\n",
    "hardest = [\"rp\", \"fr\", \"sw\"]\n",
    "easyest = [\"cf\", \"cb\", \"cm\", \"pl\", \"cd\", \"cp\", \"sw\"]\n",
    "all_ = binary_support + binary_not_support"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "respostas = {\n",
    "    'cb': [\"I\", \"H\"],\n",
    "    'cd': [\"D\", \"G\"],\n",
    "    'cf': [\"M\", \"N\"],\n",
    "    'cm': [\"I\"], # descoberto 'na mão'\n",
    "    'cp': [\"D\", \"G\"],\n",
    "    # 'fr': [\"A\", \"D\", \"G\"],\n",
    "    'fr': ['M', 'N'],\n",
    "    'lp': [\"A\", \"C\", \"D\", \"G\"],\n",
    "    'pl': [\"A\", \"D\", \"G\"],\n",
    "    'pm': [\"I\"],\n",
    "    're': [\"C\"],\n",
    "    'rp': [\"O\", \"Replaced_Activity\"], # + Replaced Activity \n",
    "    'sw': [\"O\", \"I\", \"H\"],\n",
    "    'IOR': [\"K\"], # + Added Activity\n",
    "    'IRO': [\"O\", \"I\"], # + Added Activity\n",
    "    'OIR': [\"D\", \"G\", \"A\"], # + Added Activity\n",
    "    'ORI': [\"H\", \"N\", \"M\", \"O\"], # + Added Activity\n",
    "    'RIO': [\"N\", \"I\", \"H\", \"M\"],\n",
    "    'ROI': [\"A\", \"G\"]\n",
    "}\n",
    "\n",
    "transitions_requirement = {\n",
    "    'cb': [\"C I\", \"C N\", \"C M\", 'I H', 'H N', 'H M'],\n",
    "    'cd': ['G D', 'A C'],\n",
    "    'cf': ['H N', 'H M', 'M N', 'M O'],\n",
    "    'cm': ['C H', 'C I', \"I H\"],\n",
    "    'cp': ['O G', 'O E', 'O B'],\n",
    "    # 'fr': ['F G', 'G D', 'F A', 'A C', 'D C'],\n",
    "    'fr': ['H M', 'H N', 'M O', 'N O'],\n",
    "    'lp': ['C G', 'C A', 'G D', 'D C', 'A C'],\n",
    "    'pl': ['F A', 'A G', 'G D', 'D C', 'A D'],\n",
    "    'pm': ['C I', 'C H', 'I F', 'I N', 'N I', 'N O', 'I O', 'H I'],\n",
    "    're': ['D C', 'A C', 'D I', 'D K', 'A I', \"A K\"],\n",
    "    'rp': ['N O', 'M O', 'O E', 'O B', 'N Replaced_Activity', 'M Replaced_Activity', 'Replaced_Activity E', 'Replaced_Activity B'],\n",
    "    'sw': ['C O', 'C I', 'O N', 'O M', 'M I', 'N I', 'H E', 'H B', 'H N', 'H M'] \n",
    "}"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Getting colnames and y_true for all cases"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "def insensitive_glob(pattern):\n",
    "    def either(c):\n",
    "        return '[%s%s]' % (c.lower(), c.upper()) if c.isalpha() else c\n",
    "    return glob.glob(''.join(map(either, pattern)))\n",
    "\n",
    "logs = insensitive_glob(\"process_mining_datasets/*/*k.MXML\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "def get_colnames(tipo_mudanca, log_size, aliases, representation):\n",
    "    final_name = tipo_mudanca + (str(log_size/1000)).replace(\".0\", \"\") + \"k\"\n",
    "    \n",
    "    log_file = [x for x in logs if final_name in x][0]\n",
    "    \n",
    "    log_read = pm.all_prep(log_file, aliases)\n",
    "    df = objects[\"representation\"][representation](\n",
    "        log_read\n",
    "    )\n",
    "    \n",
    "    col_names = sorted(df.columns)\n",
    "    \n",
    "    if representation == 'transitions':\n",
    "        dic = transitions_requirement\n",
    "    else:\n",
    "        dic = respostas\n",
    "    \n",
    "    resp = []\n",
    "    for col in col_names:\n",
    "        add = False\n",
    "        for act in dic[tipo_mudanca]:\n",
    "            if act in col and \"END\" not in col and \"START\" not in col:\n",
    "                add = True\n",
    "                break\n",
    "        if add:\n",
    "            resp.append(1)\n",
    "        else:\n",
    "            resp.append(0)\n",
    "    \n",
    "    return col_names, resp"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "fr 5000 transitions\n"
     ]
    },
    {
     "ename": "NameError",
     "evalue": "name 'cached_resp' is not defined",
     "output_type": "error",
     "traceback": [
      "\u001b[1;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[1;31mNameError\u001b[0m                                 Traceback (most recent call last)",
      "\u001b[1;32m<ipython-input-20-5194f1145209>\u001b[0m in \u001b[0;36m<module>\u001b[1;34m\u001b[0m\n\u001b[0;32m      5\u001b[0m         \u001b[1;32mfor\u001b[0m \u001b[0mrep\u001b[0m \u001b[1;32min\u001b[0m \u001b[1;33m[\u001b[0m\u001b[1;34m'transitions'\u001b[0m\u001b[1;33m,\u001b[0m \u001b[1;34m'binary'\u001b[0m\u001b[1;33m,\u001b[0m \u001b[1;34m'frequency'\u001b[0m\u001b[1;33m]\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m      6\u001b[0m             \u001b[0mprint\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mtipo_mudanca\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mlog_size\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mrep\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m----> 7\u001b[1;33m             \u001b[0mcached_resp\u001b[0m\u001b[1;33m[\u001b[0m\u001b[0mtipo_mudanca\u001b[0m \u001b[1;33m+\u001b[0m \u001b[1;34m\"_\"\u001b[0m \u001b[1;33m+\u001b[0m \u001b[0mstr\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mlog_size\u001b[0m\u001b[1;33m)\u001b[0m \u001b[1;33m+\u001b[0m \u001b[1;34m\"_\"\u001b[0m \u001b[1;33m+\u001b[0m \u001b[0mrep\u001b[0m\u001b[1;33m]\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mget_colnames\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mtipo_mudanca\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mlog_size\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0maliases\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mrep\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m",
      "\u001b[1;31mNameError\u001b[0m: name 'cached_resp' is not defined"
     ]
    }
   ],
   "source": [
    "# cached_resp = {}\n",
    "\n",
    "for tipo_mudanca in ['fr']: # all_ + ['re', 'lp']:\n",
    "    for log_size in [5000, 7500, 10000]: \n",
    "        for rep in ['transitions', 'binary', 'frequency']:\n",
    "            print(tipo_mudanca, log_size, rep)\n",
    "            cached_resp[tipo_mudanca + \"_\" + str(log_size) + \"_\" + rep] = get_colnames(tipo_mudanca, log_size, aliases, rep)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "json.dump(cached_resp, open(\"cached_resp_v2.json\", 'w'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "cached_resp = json.load(open(\"cached_resp_v2.json\", 'r'))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Test localization experiments"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "combinations = list(\n",
    "    ParameterGrid({\n",
    "        'tipo_mudanca': ['cp'],\n",
    "        'log_size': [5000],\n",
    "        'representation': ['frequency'],\n",
    "        'window_size': [100],\n",
    "        'model': ['kmeans__k=2'],\n",
    "        'sliding_window': [False]\n",
    "    })\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "df = pd.read_csv(base_dir + combinations[0][\"tipo_mudanca\"] + \"/\" + opts_to_string(combinations[0]) + \".csv\", index_col=0)\n",
    "df.centroids = df.centroids.apply(string_to_numpy)\n",
    "df.index += df.index[1]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "def localize_drift(old_centroids, new_centroids, col_names):\n",
    "    if isinstance(old_centroids, pd.Series):\n",
    "        old_centroids = old_centroids.mean(axis=0)\n",
    "    \n",
    "    if isinstance(new_centroids, pd.Series):\n",
    "        new_centroids = new_centroids.mean(axis=0)\n",
    "        \n",
    "    # lut = distance.cdist(old_centroids, new_centroids).argmin(axis=0)    \n",
    "    \n",
    "    r, c = linear_sum_assignment(distance.cdist(old_centroids, new_centroids))\n",
    "    lut = c\n",
    "    \n",
    "    resp = pd.DataFrame([\n",
    "        ((old_centroids - new_centroids[lut]) ** 2).mean(axis=0)\n",
    "    ], columns=col_names).transpose()\n",
    "    \n",
    "    return resp"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [],
   "source": [
    "local = localize_drift(df.centroids.loc[500], df.centroids.loc[700], cached_resp[\"cp_5000_frequency\"][0])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>0</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>G</th>\n",
       "      <td>0.500000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>D</th>\n",
       "      <td>0.500000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>F</th>\n",
       "      <td>0.018235</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>L</th>\n",
       "      <td>0.018235</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>J</th>\n",
       "      <td>0.018235</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>N</th>\n",
       "      <td>0.013889</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>M</th>\n",
       "      <td>0.013889</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>E</th>\n",
       "      <td>0.000536</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>END_A</th>\n",
       "      <td>0.000536</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>END_C</th>\n",
       "      <td>0.000536</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>B</th>\n",
       "      <td>0.000536</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>K</th>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>O</th>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>A</th>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>I</th>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>H</th>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>END_R</th>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>C</th>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>START</th>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "              0\n",
       "G      0.500000\n",
       "D      0.500000\n",
       "F      0.018235\n",
       "L      0.018235\n",
       "J      0.018235\n",
       "N      0.013889\n",
       "M      0.013889\n",
       "E      0.000536\n",
       "END_A  0.000536\n",
       "END_C  0.000536\n",
       "B      0.000536\n",
       "K      0.000000\n",
       "O      0.000000\n",
       "A      0.000000\n",
       "I      0.000000\n",
       "H      0.000000\n",
       "END_R  0.000000\n",
       "C      0.000000\n",
       "START  0.000000"
      ]
     },
     "execution_count": 19,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "local.sort_values(0, ascending=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "1.0"
      ]
     },
     "execution_count": 25,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "roc_auc_score(cached_resp[\"cp_5000_frequency\"][1], local.values.reshape(1,-1)[0])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "1.0"
      ]
     },
     "execution_count": 26,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "average_precision_score(cached_resp[\"cp_5000_frequency\"][1], local.values.reshape(1,-1)[0])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Rodagem experimentos"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [],
   "source": [
    "combinations = list(\n",
    "    ParameterGrid({\n",
    "        'tipo_mudanca': all_ + ['re', 'lp'],\n",
    "        'log_size': [10000, 5000, 7500],\n",
    "        'representation': ['transitions', 'frequency', 'binary'],\n",
    "        'window_size': [75, 100, 125, 150, 200, 250],\n",
    "        'model': ['kmeans__k=2', 'kmeans__k=3'],\n",
    "        'sliding_window': [False]\n",
    "    })\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "for i in range(len(resp__2)):\n",
    "    if resp__2[i]['tipo_mudanca'] == \"fr\":\n",
    "        del resp__2[i]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "cf 10000 transitions kmeans__k=2 75\n",
      "cf 10000 transitions kmeans__k=2 100\n",
      "cf 10000 transitions kmeans__k=2 125\n",
      "cf 10000 transitions kmeans__k=2 150\n",
      "cf 10000 transitions kmeans__k=2 200\n",
      "cf 10000 transitions kmeans__k=2 250\n",
      "cb 10000 transitions kmeans__k=2 75\n",
      "cb 10000 transitions kmeans__k=2 100\n",
      "cb 10000 transitions kmeans__k=2 125\n",
      "cb 10000 transitions kmeans__k=2 150\n",
      "cb 10000 transitions kmeans__k=2 200\n",
      "cb 10000 transitions kmeans__k=2 250\n",
      "cm 10000 transitions kmeans__k=2 75\n",
      "cm 10000 transitions kmeans__k=2 100\n",
      "cm 10000 transitions kmeans__k=2 125\n",
      "cm 10000 transitions kmeans__k=2 150\n",
      "cm 10000 transitions kmeans__k=2 200\n",
      "cm 10000 transitions kmeans__k=2 250\n",
      "rp 10000 transitions kmeans__k=2 75\n",
      "rp 10000 transitions kmeans__k=2 100\n",
      "rp 10000 transitions kmeans__k=2 125\n",
      "rp 10000 transitions kmeans__k=2 150\n",
      "rp 10000 transitions kmeans__k=2 200\n",
      "rp 10000 transitions kmeans__k=2 250\n",
      "pm 10000 transitions kmeans__k=2 75\n",
      "pm 10000 transitions kmeans__k=2 100\n",
      "pm 10000 transitions kmeans__k=2 125\n",
      "pm 10000 transitions kmeans__k=2 150\n",
      "pm 10000 transitions kmeans__k=2 200\n",
      "pm 10000 transitions kmeans__k=2 250\n",
      "fr 10000 transitions kmeans__k=2 75\n",
      "fr 10000 transitions kmeans__k=2 100\n",
      "fr 10000 transitions kmeans__k=2 125\n",
      "fr 10000 transitions kmeans__k=2 150\n",
      "fr 10000 transitions kmeans__k=2 200\n",
      "fr 10000 transitions kmeans__k=2 250\n",
      "pl 10000 transitions kmeans__k=2 75\n",
      "pl 10000 transitions kmeans__k=2 100\n",
      "pl 10000 transitions kmeans__k=2 125\n",
      "pl 10000 transitions kmeans__k=2 150\n",
      "pl 10000 transitions kmeans__k=2 200\n",
      "pl 10000 transitions kmeans__k=2 250\n",
      "cd 10000 transitions kmeans__k=2 75\n",
      "cd 10000 transitions kmeans__k=2 100\n",
      "cd 10000 transitions kmeans__k=2 125\n",
      "cd 10000 transitions kmeans__k=2 150\n",
      "cd 10000 transitions kmeans__k=2 200\n",
      "cd 10000 transitions kmeans__k=2 250\n",
      "cp 10000 transitions kmeans__k=2 75\n",
      "cp 10000 transitions kmeans__k=2 100\n",
      "cp 10000 transitions kmeans__k=2 125\n",
      "cp 10000 transitions kmeans__k=2 150\n",
      "cp 10000 transitions kmeans__k=2 200\n",
      "cp 10000 transitions kmeans__k=2 250\n",
      "sw 10000 transitions kmeans__k=2 75\n",
      "sw 10000 transitions kmeans__k=2 100\n",
      "sw 10000 transitions kmeans__k=2 125\n",
      "sw 10000 transitions kmeans__k=2 150\n",
      "sw 10000 transitions kmeans__k=2 200\n",
      "sw 10000 transitions kmeans__k=2 250\n",
      "re 10000 transitions kmeans__k=2 75\n",
      "re 10000 transitions kmeans__k=2 100\n",
      "re 10000 transitions kmeans__k=2 125\n",
      "re 10000 transitions kmeans__k=2 150\n",
      "re 10000 transitions kmeans__k=2 200\n",
      "re 10000 transitions kmeans__k=2 250\n",
      "lp 10000 transitions kmeans__k=2 75\n",
      "lp 10000 transitions kmeans__k=2 100\n",
      "lp 10000 transitions kmeans__k=2 125\n",
      "lp 10000 transitions kmeans__k=2 150\n",
      "lp 10000 transitions kmeans__k=2 200\n",
      "lp 10000 transitions kmeans__k=2 250\n",
      "cf 10000 frequency kmeans__k=2 75\n",
      "cf 10000 frequency kmeans__k=2 100\n",
      "cf 10000 frequency kmeans__k=2 125\n",
      "cf 10000 frequency kmeans__k=2 150\n",
      "cf 10000 frequency kmeans__k=2 200\n",
      "cf 10000 frequency kmeans__k=2 250\n",
      "cb 10000 frequency kmeans__k=2 75\n",
      "cb 10000 frequency kmeans__k=2 100\n",
      "cb 10000 frequency kmeans__k=2 125\n",
      "cb 10000 frequency kmeans__k=2 150\n",
      "cb 10000 frequency kmeans__k=2 200\n",
      "cb 10000 frequency kmeans__k=2 250\n",
      "cm 10000 frequency kmeans__k=2 75\n",
      "cm 10000 frequency kmeans__k=2 100\n",
      "cm 10000 frequency kmeans__k=2 125\n",
      "cm 10000 frequency kmeans__k=2 150\n",
      "cm 10000 frequency kmeans__k=2 200\n",
      "cm 10000 frequency kmeans__k=2 250\n",
      "rp 10000 frequency kmeans__k=2 75\n",
      "rp 10000 frequency kmeans__k=2 100\n",
      "rp 10000 frequency kmeans__k=2 125\n",
      "rp 10000 frequency kmeans__k=2 150\n",
      "rp 10000 frequency kmeans__k=2 200\n",
      "rp 10000 frequency kmeans__k=2 250\n",
      "pm 10000 frequency kmeans__k=2 75\n",
      "pm 10000 frequency kmeans__k=2 100\n",
      "pm 10000 frequency kmeans__k=2 125\n",
      "pm 10000 frequency kmeans__k=2 150\n",
      "pm 10000 frequency kmeans__k=2 200\n",
      "pm 10000 frequency kmeans__k=2 250\n",
      "fr 10000 frequency kmeans__k=2 75\n",
      "fr 10000 frequency kmeans__k=2 100\n",
      "fr 10000 frequency kmeans__k=2 125\n",
      "fr 10000 frequency kmeans__k=2 150\n",
      "fr 10000 frequency kmeans__k=2 200\n",
      "fr 10000 frequency kmeans__k=2 250\n",
      "pl 10000 frequency kmeans__k=2 75\n",
      "pl 10000 frequency kmeans__k=2 100\n",
      "pl 10000 frequency kmeans__k=2 125\n",
      "pl 10000 frequency kmeans__k=2 150\n",
      "pl 10000 frequency kmeans__k=2 200\n",
      "pl 10000 frequency kmeans__k=2 250\n",
      "cd 10000 frequency kmeans__k=2 75\n",
      "cd 10000 frequency kmeans__k=2 100\n",
      "cd 10000 frequency kmeans__k=2 125\n",
      "cd 10000 frequency kmeans__k=2 150\n",
      "cd 10000 frequency kmeans__k=2 200\n",
      "cd 10000 frequency kmeans__k=2 250\n",
      "cp 10000 frequency kmeans__k=2 75\n",
      "cp 10000 frequency kmeans__k=2 100\n",
      "cp 10000 frequency kmeans__k=2 125\n",
      "cp 10000 frequency kmeans__k=2 150\n",
      "cp 10000 frequency kmeans__k=2 200\n",
      "cp 10000 frequency kmeans__k=2 250\n",
      "sw 10000 frequency kmeans__k=2 75\n",
      "sw 10000 frequency kmeans__k=2 100\n",
      "sw 10000 frequency kmeans__k=2 125\n",
      "sw 10000 frequency kmeans__k=2 150\n",
      "sw 10000 frequency kmeans__k=2 200\n",
      "sw 10000 frequency kmeans__k=2 250\n",
      "re 10000 frequency kmeans__k=2 75\n",
      "re 10000 frequency kmeans__k=2 100\n",
      "re 10000 frequency kmeans__k=2 125\n",
      "re 10000 frequency kmeans__k=2 150\n",
      "re 10000 frequency kmeans__k=2 200\n",
      "re 10000 frequency kmeans__k=2 250\n",
      "lp 10000 frequency kmeans__k=2 75\n",
      "lp 10000 frequency kmeans__k=2 100\n",
      "lp 10000 frequency kmeans__k=2 125\n",
      "lp 10000 frequency kmeans__k=2 150\n",
      "lp 10000 frequency kmeans__k=2 200\n",
      "lp 10000 frequency kmeans__k=2 250\n",
      "cf 10000 binary kmeans__k=2 75\n",
      "cf 10000 binary kmeans__k=2 100\n",
      "cf 10000 binary kmeans__k=2 125\n",
      "cf 10000 binary kmeans__k=2 150\n",
      "cf 10000 binary kmeans__k=2 200\n",
      "cf 10000 binary kmeans__k=2 250\n",
      "cb 10000 binary kmeans__k=2 75\n",
      "cb 10000 binary kmeans__k=2 100\n",
      "cb 10000 binary kmeans__k=2 125\n",
      "cb 10000 binary kmeans__k=2 150\n",
      "cb 10000 binary kmeans__k=2 200\n",
      "cb 10000 binary kmeans__k=2 250\n",
      "cm 10000 binary kmeans__k=2 75\n",
      "cm 10000 binary kmeans__k=2 100\n",
      "cm 10000 binary kmeans__k=2 125\n",
      "cm 10000 binary kmeans__k=2 150\n",
      "cm 10000 binary kmeans__k=2 200\n",
      "cm 10000 binary kmeans__k=2 250\n",
      "rp 10000 binary kmeans__k=2 75\n",
      "rp 10000 binary kmeans__k=2 100\n",
      "rp 10000 binary kmeans__k=2 125\n",
      "rp 10000 binary kmeans__k=2 150\n",
      "rp 10000 binary kmeans__k=2 200\n",
      "rp 10000 binary kmeans__k=2 250\n",
      "pm 10000 binary kmeans__k=2 75\n",
      "pm 10000 binary kmeans__k=2 100\n",
      "pm 10000 binary kmeans__k=2 125\n",
      "pm 10000 binary kmeans__k=2 150\n",
      "pm 10000 binary kmeans__k=2 200\n",
      "pm 10000 binary kmeans__k=2 250\n",
      "fr 10000 binary kmeans__k=2 75\n",
      "fr 10000 binary kmeans__k=2 100\n",
      "fr 10000 binary kmeans__k=2 125\n",
      "fr 10000 binary kmeans__k=2 150\n",
      "fr 10000 binary kmeans__k=2 200\n",
      "fr 10000 binary kmeans__k=2 250\n",
      "pl 10000 binary kmeans__k=2 75\n",
      "pl 10000 binary kmeans__k=2 100\n",
      "pl 10000 binary kmeans__k=2 125\n",
      "pl 10000 binary kmeans__k=2 150\n",
      "pl 10000 binary kmeans__k=2 200\n",
      "pl 10000 binary kmeans__k=2 250\n",
      "cd 10000 binary kmeans__k=2 75\n",
      "cd 10000 binary kmeans__k=2 100\n",
      "cd 10000 binary kmeans__k=2 125\n",
      "cd 10000 binary kmeans__k=2 150\n",
      "cd 10000 binary kmeans__k=2 200\n",
      "cd 10000 binary kmeans__k=2 250\n",
      "cp 10000 binary kmeans__k=2 75\n",
      "cp 10000 binary kmeans__k=2 100\n",
      "cp 10000 binary kmeans__k=2 125\n",
      "cp 10000 binary kmeans__k=2 150\n",
      "cp 10000 binary kmeans__k=2 200\n",
      "cp 10000 binary kmeans__k=2 250\n",
      "sw 10000 binary kmeans__k=2 75\n",
      "sw 10000 binary kmeans__k=2 100\n",
      "sw 10000 binary kmeans__k=2 125\n",
      "sw 10000 binary kmeans__k=2 150\n",
      "sw 10000 binary kmeans__k=2 200\n",
      "sw 10000 binary kmeans__k=2 250\n",
      "re 10000 binary kmeans__k=2 75\n",
      "re 10000 binary kmeans__k=2 100\n",
      "re 10000 binary kmeans__k=2 125\n",
      "re 10000 binary kmeans__k=2 150\n",
      "re 10000 binary kmeans__k=2 200\n",
      "re 10000 binary kmeans__k=2 250\n",
      "lp 10000 binary kmeans__k=2 75\n",
      "lp 10000 binary kmeans__k=2 100\n",
      "lp 10000 binary kmeans__k=2 125\n",
      "lp 10000 binary kmeans__k=2 150\n",
      "lp 10000 binary kmeans__k=2 200\n",
      "lp 10000 binary kmeans__k=2 250\n",
      "cf 10000 transitions kmeans__k=3 75\n",
      "cf 10000 transitions kmeans__k=3 100\n",
      "cf 10000 transitions kmeans__k=3 125\n",
      "cf 10000 transitions kmeans__k=3 150\n",
      "cf 10000 transitions kmeans__k=3 200\n",
      "cf 10000 transitions kmeans__k=3 250\n",
      "cb 10000 transitions kmeans__k=3 75\n",
      "cb 10000 transitions kmeans__k=3 100\n",
      "cb 10000 transitions kmeans__k=3 125\n",
      "cb 10000 transitions kmeans__k=3 150\n",
      "cb 10000 transitions kmeans__k=3 200\n",
      "cb 10000 transitions kmeans__k=3 250\n",
      "cm 10000 transitions kmeans__k=3 75\n",
      "cm 10000 transitions kmeans__k=3 100\n",
      "cm 10000 transitions kmeans__k=3 125\n",
      "cm 10000 transitions kmeans__k=3 150\n",
      "cm 10000 transitions kmeans__k=3 200\n",
      "cm 10000 transitions kmeans__k=3 250\n",
      "rp 10000 transitions kmeans__k=3 75\n",
      "rp 10000 transitions kmeans__k=3 100\n",
      "rp 10000 transitions kmeans__k=3 125\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "rp 10000 transitions kmeans__k=3 150\n",
      "rp 10000 transitions kmeans__k=3 200\n",
      "rp 10000 transitions kmeans__k=3 250\n",
      "pm 10000 transitions kmeans__k=3 75\n",
      "pm 10000 transitions kmeans__k=3 100\n",
      "pm 10000 transitions kmeans__k=3 125\n",
      "pm 10000 transitions kmeans__k=3 150\n",
      "pm 10000 transitions kmeans__k=3 200\n",
      "pm 10000 transitions kmeans__k=3 250\n",
      "fr 10000 transitions kmeans__k=3 75\n",
      "fr 10000 transitions kmeans__k=3 100\n",
      "fr 10000 transitions kmeans__k=3 125\n",
      "fr 10000 transitions kmeans__k=3 150\n",
      "fr 10000 transitions kmeans__k=3 200\n",
      "fr 10000 transitions kmeans__k=3 250\n",
      "pl 10000 transitions kmeans__k=3 75\n",
      "pl 10000 transitions kmeans__k=3 100\n",
      "pl 10000 transitions kmeans__k=3 125\n",
      "pl 10000 transitions kmeans__k=3 150\n",
      "pl 10000 transitions kmeans__k=3 200\n",
      "pl 10000 transitions kmeans__k=3 250\n",
      "cd 10000 transitions kmeans__k=3 75\n",
      "cd 10000 transitions kmeans__k=3 100\n",
      "cd 10000 transitions kmeans__k=3 125\n",
      "cd 10000 transitions kmeans__k=3 150\n",
      "cd 10000 transitions kmeans__k=3 200\n",
      "cd 10000 transitions kmeans__k=3 250\n",
      "cp 10000 transitions kmeans__k=3 75\n",
      "cp 10000 transitions kmeans__k=3 100\n",
      "cp 10000 transitions kmeans__k=3 125\n",
      "cp 10000 transitions kmeans__k=3 150\n",
      "cp 10000 transitions kmeans__k=3 200\n",
      "cp 10000 transitions kmeans__k=3 250\n",
      "sw 10000 transitions kmeans__k=3 75\n",
      "sw 10000 transitions kmeans__k=3 100\n",
      "sw 10000 transitions kmeans__k=3 125\n",
      "sw 10000 transitions kmeans__k=3 150\n",
      "sw 10000 transitions kmeans__k=3 200\n",
      "sw 10000 transitions kmeans__k=3 250\n",
      "re 10000 transitions kmeans__k=3 75\n",
      "re 10000 transitions kmeans__k=3 100\n",
      "re 10000 transitions kmeans__k=3 125\n",
      "re 10000 transitions kmeans__k=3 150\n",
      "re 10000 transitions kmeans__k=3 200\n",
      "re 10000 transitions kmeans__k=3 250\n",
      "lp 10000 transitions kmeans__k=3 75\n",
      "lp 10000 transitions kmeans__k=3 100\n",
      "lp 10000 transitions kmeans__k=3 125\n",
      "lp 10000 transitions kmeans__k=3 150\n",
      "lp 10000 transitions kmeans__k=3 200\n",
      "lp 10000 transitions kmeans__k=3 250\n",
      "cf 10000 frequency kmeans__k=3 75\n",
      "cf 10000 frequency kmeans__k=3 100\n",
      "cf 10000 frequency kmeans__k=3 125\n",
      "cf 10000 frequency kmeans__k=3 150\n",
      "cf 10000 frequency kmeans__k=3 200\n",
      "cf 10000 frequency kmeans__k=3 250\n",
      "cb 10000 frequency kmeans__k=3 75\n",
      "cb 10000 frequency kmeans__k=3 100\n",
      "cb 10000 frequency kmeans__k=3 125\n",
      "cb 10000 frequency kmeans__k=3 150\n",
      "cb 10000 frequency kmeans__k=3 200\n",
      "cb 10000 frequency kmeans__k=3 250\n",
      "cm 10000 frequency kmeans__k=3 75\n",
      "cm 10000 frequency kmeans__k=3 100\n",
      "cm 10000 frequency kmeans__k=3 125\n",
      "cm 10000 frequency kmeans__k=3 150\n",
      "cm 10000 frequency kmeans__k=3 200\n",
      "cm 10000 frequency kmeans__k=3 250\n",
      "rp 10000 frequency kmeans__k=3 75\n",
      "rp 10000 frequency kmeans__k=3 100\n",
      "rp 10000 frequency kmeans__k=3 125\n",
      "rp 10000 frequency kmeans__k=3 150\n",
      "rp 10000 frequency kmeans__k=3 200\n",
      "rp 10000 frequency kmeans__k=3 250\n",
      "pm 10000 frequency kmeans__k=3 75\n",
      "pm 10000 frequency kmeans__k=3 100\n",
      "pm 10000 frequency kmeans__k=3 125\n",
      "pm 10000 frequency kmeans__k=3 150\n",
      "pm 10000 frequency kmeans__k=3 200\n",
      "pm 10000 frequency kmeans__k=3 250\n",
      "fr 10000 frequency kmeans__k=3 75\n",
      "fr 10000 frequency kmeans__k=3 100\n",
      "fr 10000 frequency kmeans__k=3 125\n",
      "fr 10000 frequency kmeans__k=3 150\n",
      "fr 10000 frequency kmeans__k=3 200\n",
      "fr 10000 frequency kmeans__k=3 250\n",
      "pl 10000 frequency kmeans__k=3 75\n",
      "pl 10000 frequency kmeans__k=3 100\n",
      "pl 10000 frequency kmeans__k=3 125\n",
      "pl 10000 frequency kmeans__k=3 150\n",
      "pl 10000 frequency kmeans__k=3 200\n",
      "pl 10000 frequency kmeans__k=3 250\n",
      "cd 10000 frequency kmeans__k=3 75\n",
      "cd 10000 frequency kmeans__k=3 100\n",
      "cd 10000 frequency kmeans__k=3 125\n",
      "cd 10000 frequency kmeans__k=3 150\n",
      "cd 10000 frequency kmeans__k=3 200\n",
      "cd 10000 frequency kmeans__k=3 250\n",
      "cp 10000 frequency kmeans__k=3 75\n",
      "cp 10000 frequency kmeans__k=3 100\n",
      "cp 10000 frequency kmeans__k=3 125\n",
      "cp 10000 frequency kmeans__k=3 150\n",
      "cp 10000 frequency kmeans__k=3 200\n",
      "cp 10000 frequency kmeans__k=3 250\n",
      "sw 10000 frequency kmeans__k=3 75\n",
      "sw 10000 frequency kmeans__k=3 100\n",
      "sw 10000 frequency kmeans__k=3 125\n",
      "sw 10000 frequency kmeans__k=3 150\n",
      "sw 10000 frequency kmeans__k=3 200\n",
      "sw 10000 frequency kmeans__k=3 250\n",
      "re 10000 frequency kmeans__k=3 75\n",
      "re 10000 frequency kmeans__k=3 100\n",
      "re 10000 frequency kmeans__k=3 125\n",
      "re 10000 frequency kmeans__k=3 150\n",
      "re 10000 frequency kmeans__k=3 200\n",
      "re 10000 frequency kmeans__k=3 250\n",
      "lp 10000 frequency kmeans__k=3 75\n",
      "lp 10000 frequency kmeans__k=3 100\n",
      "lp 10000 frequency kmeans__k=3 125\n",
      "lp 10000 frequency kmeans__k=3 150\n",
      "lp 10000 frequency kmeans__k=3 200\n",
      "lp 10000 frequency kmeans__k=3 250\n",
      "cf 10000 binary kmeans__k=3 75\n",
      "cf 10000 binary kmeans__k=3 100\n",
      "cf 10000 binary kmeans__k=3 125\n",
      "cf 10000 binary kmeans__k=3 150\n",
      "cf 10000 binary kmeans__k=3 200\n",
      "cf 10000 binary kmeans__k=3 250\n",
      "cb 10000 binary kmeans__k=3 75\n",
      "cb 10000 binary kmeans__k=3 100\n",
      "cb 10000 binary kmeans__k=3 125\n",
      "cb 10000 binary kmeans__k=3 150\n",
      "cb 10000 binary kmeans__k=3 200\n",
      "cb 10000 binary kmeans__k=3 250\n",
      "cm 10000 binary kmeans__k=3 75\n",
      "cm 10000 binary kmeans__k=3 100\n",
      "cm 10000 binary kmeans__k=3 125\n",
      "cm 10000 binary kmeans__k=3 150\n",
      "cm 10000 binary kmeans__k=3 200\n",
      "cm 10000 binary kmeans__k=3 250\n",
      "rp 10000 binary kmeans__k=3 75\n",
      "rp 10000 binary kmeans__k=3 100\n",
      "rp 10000 binary kmeans__k=3 125\n",
      "rp 10000 binary kmeans__k=3 150\n",
      "rp 10000 binary kmeans__k=3 200\n",
      "rp 10000 binary kmeans__k=3 250\n",
      "pm 10000 binary kmeans__k=3 75\n",
      "pm 10000 binary kmeans__k=3 100\n",
      "pm 10000 binary kmeans__k=3 125\n",
      "pm 10000 binary kmeans__k=3 150\n",
      "pm 10000 binary kmeans__k=3 200\n",
      "pm 10000 binary kmeans__k=3 250\n",
      "fr 10000 binary kmeans__k=3 75\n",
      "fr 10000 binary kmeans__k=3 100\n",
      "fr 10000 binary kmeans__k=3 125\n",
      "fr 10000 binary kmeans__k=3 150\n",
      "fr 10000 binary kmeans__k=3 200\n",
      "fr 10000 binary kmeans__k=3 250\n",
      "pl 10000 binary kmeans__k=3 75\n",
      "pl 10000 binary kmeans__k=3 100\n",
      "pl 10000 binary kmeans__k=3 125\n",
      "pl 10000 binary kmeans__k=3 150\n",
      "pl 10000 binary kmeans__k=3 200\n",
      "pl 10000 binary kmeans__k=3 250\n",
      "cd 10000 binary kmeans__k=3 75\n",
      "cd 10000 binary kmeans__k=3 100\n",
      "cd 10000 binary kmeans__k=3 125\n",
      "cd 10000 binary kmeans__k=3 150\n",
      "cd 10000 binary kmeans__k=3 200\n",
      "cd 10000 binary kmeans__k=3 250\n",
      "cp 10000 binary kmeans__k=3 75\n",
      "cp 10000 binary kmeans__k=3 100\n",
      "cp 10000 binary kmeans__k=3 125\n",
      "cp 10000 binary kmeans__k=3 150\n",
      "cp 10000 binary kmeans__k=3 200\n",
      "cp 10000 binary kmeans__k=3 250\n",
      "sw 10000 binary kmeans__k=3 75\n",
      "sw 10000 binary kmeans__k=3 100\n",
      "sw 10000 binary kmeans__k=3 125\n",
      "sw 10000 binary kmeans__k=3 150\n",
      "sw 10000 binary kmeans__k=3 200\n",
      "sw 10000 binary kmeans__k=3 250\n",
      "re 10000 binary kmeans__k=3 75\n",
      "re 10000 binary kmeans__k=3 100\n",
      "re 10000 binary kmeans__k=3 125\n",
      "re 10000 binary kmeans__k=3 150\n",
      "re 10000 binary kmeans__k=3 200\n",
      "re 10000 binary kmeans__k=3 250\n",
      "lp 10000 binary kmeans__k=3 75\n",
      "lp 10000 binary kmeans__k=3 100\n",
      "lp 10000 binary kmeans__k=3 125\n",
      "lp 10000 binary kmeans__k=3 150\n",
      "lp 10000 binary kmeans__k=3 200\n",
      "lp 10000 binary kmeans__k=3 250\n",
      "cf 5000 transitions kmeans__k=2 75\n",
      "cf 5000 transitions kmeans__k=2 100\n",
      "cf 5000 transitions kmeans__k=2 125\n",
      "cf 5000 transitions kmeans__k=2 150\n",
      "cf 5000 transitions kmeans__k=2 200\n",
      "cf 5000 transitions kmeans__k=2 250\n",
      "cb 5000 transitions kmeans__k=2 75\n",
      "cb 5000 transitions kmeans__k=2 100\n",
      "cb 5000 transitions kmeans__k=2 125\n",
      "cb 5000 transitions kmeans__k=2 150\n",
      "cb 5000 transitions kmeans__k=2 200\n",
      "cb 5000 transitions kmeans__k=2 250\n",
      "cm 5000 transitions kmeans__k=2 75\n",
      "cm 5000 transitions kmeans__k=2 100\n",
      "cm 5000 transitions kmeans__k=2 125\n",
      "cm 5000 transitions kmeans__k=2 150\n",
      "cm 5000 transitions kmeans__k=2 200\n",
      "cm 5000 transitions kmeans__k=2 250\n",
      "rp 5000 transitions kmeans__k=2 75\n",
      "rp 5000 transitions kmeans__k=2 100\n",
      "rp 5000 transitions kmeans__k=2 125\n",
      "rp 5000 transitions kmeans__k=2 150\n",
      "rp 5000 transitions kmeans__k=2 200\n",
      "rp 5000 transitions kmeans__k=2 250\n",
      "pm 5000 transitions kmeans__k=2 75\n",
      "pm 5000 transitions kmeans__k=2 100\n",
      "pm 5000 transitions kmeans__k=2 125\n",
      "pm 5000 transitions kmeans__k=2 150\n",
      "pm 5000 transitions kmeans__k=2 200\n",
      "pm 5000 transitions kmeans__k=2 250\n",
      "fr 5000 transitions kmeans__k=2 75\n",
      "fr 5000 transitions kmeans__k=2 100\n",
      "fr 5000 transitions kmeans__k=2 125\n",
      "fr 5000 transitions kmeans__k=2 150\n",
      "fr 5000 transitions kmeans__k=2 200\n",
      "fr 5000 transitions kmeans__k=2 250\n",
      "pl 5000 transitions kmeans__k=2 75\n",
      "pl 5000 transitions kmeans__k=2 100\n",
      "pl 5000 transitions kmeans__k=2 125\n",
      "pl 5000 transitions kmeans__k=2 150\n",
      "pl 5000 transitions kmeans__k=2 200\n",
      "pl 5000 transitions kmeans__k=2 250\n",
      "cd 5000 transitions kmeans__k=2 75\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "cd 5000 transitions kmeans__k=2 100\n",
      "cd 5000 transitions kmeans__k=2 125\n",
      "cd 5000 transitions kmeans__k=2 150\n",
      "cd 5000 transitions kmeans__k=2 200\n",
      "cd 5000 transitions kmeans__k=2 250\n",
      "cp 5000 transitions kmeans__k=2 75\n",
      "cp 5000 transitions kmeans__k=2 100\n",
      "cp 5000 transitions kmeans__k=2 125\n",
      "cp 5000 transitions kmeans__k=2 150\n",
      "cp 5000 transitions kmeans__k=2 200\n",
      "cp 5000 transitions kmeans__k=2 250\n",
      "sw 5000 transitions kmeans__k=2 75\n",
      "sw 5000 transitions kmeans__k=2 100\n",
      "sw 5000 transitions kmeans__k=2 125\n",
      "sw 5000 transitions kmeans__k=2 150\n",
      "sw 5000 transitions kmeans__k=2 200\n",
      "sw 5000 transitions kmeans__k=2 250\n",
      "re 5000 transitions kmeans__k=2 75\n",
      "re 5000 transitions kmeans__k=2 100\n",
      "re 5000 transitions kmeans__k=2 125\n",
      "re 5000 transitions kmeans__k=2 150\n",
      "re 5000 transitions kmeans__k=2 200\n",
      "re 5000 transitions kmeans__k=2 250\n",
      "lp 5000 transitions kmeans__k=2 75\n",
      "lp 5000 transitions kmeans__k=2 100\n",
      "lp 5000 transitions kmeans__k=2 125\n",
      "lp 5000 transitions kmeans__k=2 150\n",
      "lp 5000 transitions kmeans__k=2 200\n",
      "lp 5000 transitions kmeans__k=2 250\n",
      "cf 5000 frequency kmeans__k=2 75\n",
      "cf 5000 frequency kmeans__k=2 100\n",
      "cf 5000 frequency kmeans__k=2 125\n",
      "cf 5000 frequency kmeans__k=2 150\n",
      "cf 5000 frequency kmeans__k=2 200\n",
      "cf 5000 frequency kmeans__k=2 250\n",
      "cb 5000 frequency kmeans__k=2 75\n",
      "cb 5000 frequency kmeans__k=2 100\n",
      "cb 5000 frequency kmeans__k=2 125\n",
      "cb 5000 frequency kmeans__k=2 150\n",
      "cb 5000 frequency kmeans__k=2 200\n",
      "cb 5000 frequency kmeans__k=2 250\n",
      "cm 5000 frequency kmeans__k=2 75\n",
      "cm 5000 frequency kmeans__k=2 100\n",
      "cm 5000 frequency kmeans__k=2 125\n",
      "cm 5000 frequency kmeans__k=2 150\n",
      "cm 5000 frequency kmeans__k=2 200\n",
      "cm 5000 frequency kmeans__k=2 250\n",
      "rp 5000 frequency kmeans__k=2 75\n",
      "rp 5000 frequency kmeans__k=2 100\n",
      "rp 5000 frequency kmeans__k=2 125\n",
      "rp 5000 frequency kmeans__k=2 150\n",
      "rp 5000 frequency kmeans__k=2 200\n",
      "rp 5000 frequency kmeans__k=2 250\n",
      "pm 5000 frequency kmeans__k=2 75\n",
      "pm 5000 frequency kmeans__k=2 100\n",
      "pm 5000 frequency kmeans__k=2 125\n",
      "pm 5000 frequency kmeans__k=2 150\n",
      "pm 5000 frequency kmeans__k=2 200\n",
      "pm 5000 frequency kmeans__k=2 250\n",
      "fr 5000 frequency kmeans__k=2 75\n",
      "fr 5000 frequency kmeans__k=2 100\n",
      "fr 5000 frequency kmeans__k=2 125\n",
      "fr 5000 frequency kmeans__k=2 150\n",
      "fr 5000 frequency kmeans__k=2 200\n",
      "fr 5000 frequency kmeans__k=2 250\n",
      "pl 5000 frequency kmeans__k=2 75\n",
      "pl 5000 frequency kmeans__k=2 100\n",
      "pl 5000 frequency kmeans__k=2 125\n",
      "pl 5000 frequency kmeans__k=2 150\n",
      "pl 5000 frequency kmeans__k=2 200\n",
      "pl 5000 frequency kmeans__k=2 250\n",
      "cd 5000 frequency kmeans__k=2 75\n",
      "cd 5000 frequency kmeans__k=2 100\n",
      "cd 5000 frequency kmeans__k=2 125\n",
      "cd 5000 frequency kmeans__k=2 150\n",
      "cd 5000 frequency kmeans__k=2 200\n",
      "cd 5000 frequency kmeans__k=2 250\n",
      "cp 5000 frequency kmeans__k=2 75\n",
      "cp 5000 frequency kmeans__k=2 100\n",
      "cp 5000 frequency kmeans__k=2 125\n",
      "cp 5000 frequency kmeans__k=2 150\n",
      "cp 5000 frequency kmeans__k=2 200\n",
      "cp 5000 frequency kmeans__k=2 250\n",
      "sw 5000 frequency kmeans__k=2 75\n",
      "sw 5000 frequency kmeans__k=2 100\n",
      "sw 5000 frequency kmeans__k=2 125\n",
      "sw 5000 frequency kmeans__k=2 150\n",
      "sw 5000 frequency kmeans__k=2 200\n",
      "sw 5000 frequency kmeans__k=2 250\n",
      "re 5000 frequency kmeans__k=2 75\n",
      "re 5000 frequency kmeans__k=2 100\n",
      "re 5000 frequency kmeans__k=2 125\n",
      "re 5000 frequency kmeans__k=2 150\n",
      "re 5000 frequency kmeans__k=2 200\n",
      "re 5000 frequency kmeans__k=2 250\n",
      "lp 5000 frequency kmeans__k=2 75\n",
      "lp 5000 frequency kmeans__k=2 100\n",
      "lp 5000 frequency kmeans__k=2 125\n",
      "lp 5000 frequency kmeans__k=2 150\n",
      "lp 5000 frequency kmeans__k=2 200\n",
      "lp 5000 frequency kmeans__k=2 250\n",
      "cf 5000 binary kmeans__k=2 75\n",
      "cf 5000 binary kmeans__k=2 100\n",
      "cf 5000 binary kmeans__k=2 125\n",
      "cf 5000 binary kmeans__k=2 150\n",
      "cf 5000 binary kmeans__k=2 200\n",
      "cf 5000 binary kmeans__k=2 250\n",
      "cb 5000 binary kmeans__k=2 75\n",
      "cb 5000 binary kmeans__k=2 100\n",
      "cb 5000 binary kmeans__k=2 125\n",
      "cb 5000 binary kmeans__k=2 150\n",
      "cb 5000 binary kmeans__k=2 200\n",
      "cb 5000 binary kmeans__k=2 250\n",
      "cm 5000 binary kmeans__k=2 75\n",
      "cm 5000 binary kmeans__k=2 100\n",
      "cm 5000 binary kmeans__k=2 125\n",
      "cm 5000 binary kmeans__k=2 150\n",
      "cm 5000 binary kmeans__k=2 200\n",
      "cm 5000 binary kmeans__k=2 250\n",
      "rp 5000 binary kmeans__k=2 75\n",
      "rp 5000 binary kmeans__k=2 100\n",
      "rp 5000 binary kmeans__k=2 125\n",
      "rp 5000 binary kmeans__k=2 150\n",
      "rp 5000 binary kmeans__k=2 200\n",
      "rp 5000 binary kmeans__k=2 250\n",
      "pm 5000 binary kmeans__k=2 75\n",
      "pm 5000 binary kmeans__k=2 100\n",
      "pm 5000 binary kmeans__k=2 125\n",
      "pm 5000 binary kmeans__k=2 150\n",
      "pm 5000 binary kmeans__k=2 200\n",
      "pm 5000 binary kmeans__k=2 250\n",
      "fr 5000 binary kmeans__k=2 75\n",
      "fr 5000 binary kmeans__k=2 100\n",
      "fr 5000 binary kmeans__k=2 125\n",
      "fr 5000 binary kmeans__k=2 150\n",
      "fr 5000 binary kmeans__k=2 200\n",
      "fr 5000 binary kmeans__k=2 250\n",
      "pl 5000 binary kmeans__k=2 75\n",
      "pl 5000 binary kmeans__k=2 100\n",
      "pl 5000 binary kmeans__k=2 125\n",
      "pl 5000 binary kmeans__k=2 150\n",
      "pl 5000 binary kmeans__k=2 200\n",
      "pl 5000 binary kmeans__k=2 250\n",
      "cd 5000 binary kmeans__k=2 75\n",
      "cd 5000 binary kmeans__k=2 100\n",
      "cd 5000 binary kmeans__k=2 125\n",
      "cd 5000 binary kmeans__k=2 150\n",
      "cd 5000 binary kmeans__k=2 200\n",
      "cd 5000 binary kmeans__k=2 250\n",
      "cp 5000 binary kmeans__k=2 75\n",
      "cp 5000 binary kmeans__k=2 100\n",
      "cp 5000 binary kmeans__k=2 125\n",
      "cp 5000 binary kmeans__k=2 150\n",
      "cp 5000 binary kmeans__k=2 200\n",
      "cp 5000 binary kmeans__k=2 250\n",
      "sw 5000 binary kmeans__k=2 75\n",
      "sw 5000 binary kmeans__k=2 100\n",
      "sw 5000 binary kmeans__k=2 125\n",
      "sw 5000 binary kmeans__k=2 150\n",
      "sw 5000 binary kmeans__k=2 200\n",
      "sw 5000 binary kmeans__k=2 250\n",
      "re 5000 binary kmeans__k=2 75\n",
      "re 5000 binary kmeans__k=2 100\n",
      "re 5000 binary kmeans__k=2 125\n",
      "re 5000 binary kmeans__k=2 150\n",
      "re 5000 binary kmeans__k=2 200\n",
      "re 5000 binary kmeans__k=2 250\n",
      "lp 5000 binary kmeans__k=2 75\n",
      "lp 5000 binary kmeans__k=2 100\n",
      "lp 5000 binary kmeans__k=2 125\n",
      "lp 5000 binary kmeans__k=2 150\n",
      "lp 5000 binary kmeans__k=2 200\n",
      "lp 5000 binary kmeans__k=2 250\n",
      "cf 5000 transitions kmeans__k=3 75\n",
      "cf 5000 transitions kmeans__k=3 100\n",
      "cf 5000 transitions kmeans__k=3 125\n",
      "cf 5000 transitions kmeans__k=3 150\n",
      "cf 5000 transitions kmeans__k=3 200\n",
      "cf 5000 transitions kmeans__k=3 250\n",
      "cb 5000 transitions kmeans__k=3 75\n",
      "cb 5000 transitions kmeans__k=3 100\n",
      "cb 5000 transitions kmeans__k=3 125\n",
      "cb 5000 transitions kmeans__k=3 150\n",
      "cb 5000 transitions kmeans__k=3 200\n",
      "cb 5000 transitions kmeans__k=3 250\n",
      "cm 5000 transitions kmeans__k=3 75\n",
      "cm 5000 transitions kmeans__k=3 100\n",
      "cm 5000 transitions kmeans__k=3 125\n",
      "cm 5000 transitions kmeans__k=3 150\n",
      "cm 5000 transitions kmeans__k=3 200\n",
      "cm 5000 transitions kmeans__k=3 250\n",
      "rp 5000 transitions kmeans__k=3 75\n",
      "rp 5000 transitions kmeans__k=3 100\n",
      "rp 5000 transitions kmeans__k=3 125\n",
      "rp 5000 transitions kmeans__k=3 150\n",
      "rp 5000 transitions kmeans__k=3 200\n",
      "rp 5000 transitions kmeans__k=3 250\n",
      "pm 5000 transitions kmeans__k=3 75\n",
      "pm 5000 transitions kmeans__k=3 100\n",
      "pm 5000 transitions kmeans__k=3 125\n",
      "pm 5000 transitions kmeans__k=3 150\n",
      "pm 5000 transitions kmeans__k=3 200\n",
      "pm 5000 transitions kmeans__k=3 250\n",
      "fr 5000 transitions kmeans__k=3 75\n",
      "fr 5000 transitions kmeans__k=3 100\n",
      "fr 5000 transitions kmeans__k=3 125\n",
      "fr 5000 transitions kmeans__k=3 150\n",
      "fr 5000 transitions kmeans__k=3 200\n",
      "fr 5000 transitions kmeans__k=3 250\n",
      "pl 5000 transitions kmeans__k=3 75\n",
      "pl 5000 transitions kmeans__k=3 100\n",
      "pl 5000 transitions kmeans__k=3 125\n",
      "pl 5000 transitions kmeans__k=3 150\n",
      "pl 5000 transitions kmeans__k=3 200\n",
      "pl 5000 transitions kmeans__k=3 250\n",
      "cd 5000 transitions kmeans__k=3 75\n",
      "cd 5000 transitions kmeans__k=3 100\n",
      "cd 5000 transitions kmeans__k=3 125\n",
      "cd 5000 transitions kmeans__k=3 150\n",
      "cd 5000 transitions kmeans__k=3 200\n",
      "cd 5000 transitions kmeans__k=3 250\n",
      "cp 5000 transitions kmeans__k=3 75\n",
      "cp 5000 transitions kmeans__k=3 100\n",
      "cp 5000 transitions kmeans__k=3 125\n",
      "cp 5000 transitions kmeans__k=3 150\n",
      "cp 5000 transitions kmeans__k=3 200\n",
      "cp 5000 transitions kmeans__k=3 250\n",
      "sw 5000 transitions kmeans__k=3 75\n",
      "sw 5000 transitions kmeans__k=3 100\n",
      "sw 5000 transitions kmeans__k=3 125\n",
      "sw 5000 transitions kmeans__k=3 150\n",
      "sw 5000 transitions kmeans__k=3 200\n",
      "sw 5000 transitions kmeans__k=3 250\n",
      "re 5000 transitions kmeans__k=3 75\n",
      "re 5000 transitions kmeans__k=3 100\n",
      "re 5000 transitions kmeans__k=3 125\n",
      "re 5000 transitions kmeans__k=3 150\n",
      "re 5000 transitions kmeans__k=3 200\n",
      "re 5000 transitions kmeans__k=3 250\n",
      "lp 5000 transitions kmeans__k=3 75\n",
      "lp 5000 transitions kmeans__k=3 100\n",
      "lp 5000 transitions kmeans__k=3 125\n",
      "lp 5000 transitions kmeans__k=3 150\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "lp 5000 transitions kmeans__k=3 200\n",
      "lp 5000 transitions kmeans__k=3 250\n",
      "cf 5000 frequency kmeans__k=3 75\n",
      "cf 5000 frequency kmeans__k=3 100\n",
      "cf 5000 frequency kmeans__k=3 125\n",
      "cf 5000 frequency kmeans__k=3 150\n",
      "cf 5000 frequency kmeans__k=3 200\n",
      "cf 5000 frequency kmeans__k=3 250\n",
      "cb 5000 frequency kmeans__k=3 75\n",
      "cb 5000 frequency kmeans__k=3 100\n",
      "cb 5000 frequency kmeans__k=3 125\n",
      "cb 5000 frequency kmeans__k=3 150\n",
      "cb 5000 frequency kmeans__k=3 200\n",
      "cb 5000 frequency kmeans__k=3 250\n",
      "cm 5000 frequency kmeans__k=3 75\n",
      "cm 5000 frequency kmeans__k=3 100\n",
      "cm 5000 frequency kmeans__k=3 125\n",
      "cm 5000 frequency kmeans__k=3 150\n",
      "cm 5000 frequency kmeans__k=3 200\n",
      "cm 5000 frequency kmeans__k=3 250\n",
      "rp 5000 frequency kmeans__k=3 75\n",
      "rp 5000 frequency kmeans__k=3 100\n",
      "rp 5000 frequency kmeans__k=3 125\n",
      "rp 5000 frequency kmeans__k=3 150\n",
      "rp 5000 frequency kmeans__k=3 200\n",
      "rp 5000 frequency kmeans__k=3 250\n",
      "pm 5000 frequency kmeans__k=3 75\n",
      "pm 5000 frequency kmeans__k=3 100\n",
      "pm 5000 frequency kmeans__k=3 125\n",
      "pm 5000 frequency kmeans__k=3 150\n",
      "pm 5000 frequency kmeans__k=3 200\n",
      "pm 5000 frequency kmeans__k=3 250\n",
      "fr 5000 frequency kmeans__k=3 75\n",
      "fr 5000 frequency kmeans__k=3 100\n",
      "fr 5000 frequency kmeans__k=3 125\n",
      "fr 5000 frequency kmeans__k=3 150\n",
      "fr 5000 frequency kmeans__k=3 200\n",
      "fr 5000 frequency kmeans__k=3 250\n",
      "pl 5000 frequency kmeans__k=3 75\n",
      "pl 5000 frequency kmeans__k=3 100\n",
      "pl 5000 frequency kmeans__k=3 125\n",
      "pl 5000 frequency kmeans__k=3 150\n",
      "pl 5000 frequency kmeans__k=3 200\n",
      "pl 5000 frequency kmeans__k=3 250\n",
      "cd 5000 frequency kmeans__k=3 75\n",
      "cd 5000 frequency kmeans__k=3 100\n",
      "cd 5000 frequency kmeans__k=3 125\n",
      "cd 5000 frequency kmeans__k=3 150\n",
      "cd 5000 frequency kmeans__k=3 200\n",
      "cd 5000 frequency kmeans__k=3 250\n",
      "cp 5000 frequency kmeans__k=3 75\n",
      "cp 5000 frequency kmeans__k=3 100\n",
      "cp 5000 frequency kmeans__k=3 125\n",
      "cp 5000 frequency kmeans__k=3 150\n",
      "cp 5000 frequency kmeans__k=3 200\n",
      "cp 5000 frequency kmeans__k=3 250\n",
      "sw 5000 frequency kmeans__k=3 75\n",
      "sw 5000 frequency kmeans__k=3 100\n",
      "sw 5000 frequency kmeans__k=3 125\n",
      "sw 5000 frequency kmeans__k=3 150\n",
      "sw 5000 frequency kmeans__k=3 200\n",
      "sw 5000 frequency kmeans__k=3 250\n",
      "re 5000 frequency kmeans__k=3 75\n",
      "re 5000 frequency kmeans__k=3 100\n",
      "re 5000 frequency kmeans__k=3 125\n",
      "re 5000 frequency kmeans__k=3 150\n",
      "re 5000 frequency kmeans__k=3 200\n",
      "re 5000 frequency kmeans__k=3 250\n",
      "lp 5000 frequency kmeans__k=3 75\n",
      "lp 5000 frequency kmeans__k=3 100\n",
      "lp 5000 frequency kmeans__k=3 125\n",
      "lp 5000 frequency kmeans__k=3 150\n",
      "lp 5000 frequency kmeans__k=3 200\n",
      "lp 5000 frequency kmeans__k=3 250\n",
      "cf 5000 binary kmeans__k=3 75\n",
      "cf 5000 binary kmeans__k=3 100\n",
      "cf 5000 binary kmeans__k=3 125\n",
      "cf 5000 binary kmeans__k=3 150\n",
      "cf 5000 binary kmeans__k=3 200\n",
      "cf 5000 binary kmeans__k=3 250\n",
      "cb 5000 binary kmeans__k=3 75\n",
      "cb 5000 binary kmeans__k=3 100\n",
      "cb 5000 binary kmeans__k=3 125\n",
      "cb 5000 binary kmeans__k=3 150\n",
      "cb 5000 binary kmeans__k=3 200\n",
      "cb 5000 binary kmeans__k=3 250\n",
      "cm 5000 binary kmeans__k=3 75\n",
      "cm 5000 binary kmeans__k=3 100\n",
      "cm 5000 binary kmeans__k=3 125\n",
      "cm 5000 binary kmeans__k=3 150\n",
      "cm 5000 binary kmeans__k=3 200\n",
      "cm 5000 binary kmeans__k=3 250\n",
      "rp 5000 binary kmeans__k=3 75\n",
      "rp 5000 binary kmeans__k=3 100\n",
      "rp 5000 binary kmeans__k=3 125\n",
      "rp 5000 binary kmeans__k=3 150\n",
      "rp 5000 binary kmeans__k=3 200\n",
      "rp 5000 binary kmeans__k=3 250\n",
      "pm 5000 binary kmeans__k=3 75\n",
      "pm 5000 binary kmeans__k=3 100\n",
      "pm 5000 binary kmeans__k=3 125\n",
      "pm 5000 binary kmeans__k=3 150\n",
      "pm 5000 binary kmeans__k=3 200\n",
      "pm 5000 binary kmeans__k=3 250\n",
      "fr 5000 binary kmeans__k=3 75\n",
      "fr 5000 binary kmeans__k=3 100\n",
      "fr 5000 binary kmeans__k=3 125\n",
      "fr 5000 binary kmeans__k=3 150\n",
      "fr 5000 binary kmeans__k=3 200\n",
      "fr 5000 binary kmeans__k=3 250\n",
      "pl 5000 binary kmeans__k=3 75\n",
      "pl 5000 binary kmeans__k=3 100\n",
      "pl 5000 binary kmeans__k=3 125\n",
      "pl 5000 binary kmeans__k=3 150\n",
      "pl 5000 binary kmeans__k=3 200\n",
      "pl 5000 binary kmeans__k=3 250\n",
      "cd 5000 binary kmeans__k=3 75\n",
      "cd 5000 binary kmeans__k=3 100\n",
      "cd 5000 binary kmeans__k=3 125\n",
      "cd 5000 binary kmeans__k=3 150\n",
      "cd 5000 binary kmeans__k=3 200\n",
      "cd 5000 binary kmeans__k=3 250\n",
      "cp 5000 binary kmeans__k=3 75\n",
      "cp 5000 binary kmeans__k=3 100\n",
      "cp 5000 binary kmeans__k=3 125\n",
      "cp 5000 binary kmeans__k=3 150\n",
      "cp 5000 binary kmeans__k=3 200\n",
      "cp 5000 binary kmeans__k=3 250\n",
      "sw 5000 binary kmeans__k=3 75\n",
      "sw 5000 binary kmeans__k=3 100\n",
      "sw 5000 binary kmeans__k=3 125\n",
      "sw 5000 binary kmeans__k=3 150\n",
      "sw 5000 binary kmeans__k=3 200\n",
      "sw 5000 binary kmeans__k=3 250\n",
      "re 5000 binary kmeans__k=3 75\n",
      "re 5000 binary kmeans__k=3 100\n",
      "re 5000 binary kmeans__k=3 125\n",
      "re 5000 binary kmeans__k=3 150\n",
      "re 5000 binary kmeans__k=3 200\n",
      "re 5000 binary kmeans__k=3 250\n",
      "lp 5000 binary kmeans__k=3 75\n",
      "lp 5000 binary kmeans__k=3 100\n",
      "lp 5000 binary kmeans__k=3 125\n",
      "lp 5000 binary kmeans__k=3 150\n",
      "lp 5000 binary kmeans__k=3 200\n",
      "lp 5000 binary kmeans__k=3 250\n",
      "cf 7500 transitions kmeans__k=2 75\n",
      "cf 7500 transitions kmeans__k=2 100\n",
      "cf 7500 transitions kmeans__k=2 125\n",
      "cf 7500 transitions kmeans__k=2 150\n",
      "cf 7500 transitions kmeans__k=2 200\n",
      "cf 7500 transitions kmeans__k=2 250\n",
      "cb 7500 transitions kmeans__k=2 75\n",
      "cb 7500 transitions kmeans__k=2 100\n",
      "cb 7500 transitions kmeans__k=2 125\n",
      "cb 7500 transitions kmeans__k=2 150\n",
      "cb 7500 transitions kmeans__k=2 200\n",
      "cb 7500 transitions kmeans__k=2 250\n",
      "cm 7500 transitions kmeans__k=2 75\n",
      "cm 7500 transitions kmeans__k=2 100\n",
      "cm 7500 transitions kmeans__k=2 125\n",
      "cm 7500 transitions kmeans__k=2 150\n",
      "cm 7500 transitions kmeans__k=2 200\n",
      "cm 7500 transitions kmeans__k=2 250\n",
      "rp 7500 transitions kmeans__k=2 75\n",
      "rp 7500 transitions kmeans__k=2 100\n",
      "rp 7500 transitions kmeans__k=2 125\n",
      "rp 7500 transitions kmeans__k=2 150\n",
      "rp 7500 transitions kmeans__k=2 200\n",
      "rp 7500 transitions kmeans__k=2 250\n",
      "pm 7500 transitions kmeans__k=2 75\n",
      "pm 7500 transitions kmeans__k=2 100\n",
      "pm 7500 transitions kmeans__k=2 125\n",
      "pm 7500 transitions kmeans__k=2 150\n",
      "pm 7500 transitions kmeans__k=2 200\n",
      "pm 7500 transitions kmeans__k=2 250\n",
      "fr 7500 transitions kmeans__k=2 75\n",
      "fr 7500 transitions kmeans__k=2 100\n",
      "fr 7500 transitions kmeans__k=2 125\n",
      "fr 7500 transitions kmeans__k=2 150\n",
      "fr 7500 transitions kmeans__k=2 200\n",
      "fr 7500 transitions kmeans__k=2 250\n",
      "pl 7500 transitions kmeans__k=2 75\n",
      "pl 7500 transitions kmeans__k=2 100\n",
      "pl 7500 transitions kmeans__k=2 125\n",
      "pl 7500 transitions kmeans__k=2 150\n",
      "pl 7500 transitions kmeans__k=2 200\n",
      "pl 7500 transitions kmeans__k=2 250\n",
      "cd 7500 transitions kmeans__k=2 75\n",
      "cd 7500 transitions kmeans__k=2 100\n",
      "cd 7500 transitions kmeans__k=2 125\n",
      "cd 7500 transitions kmeans__k=2 150\n",
      "cd 7500 transitions kmeans__k=2 200\n",
      "cd 7500 transitions kmeans__k=2 250\n",
      "cp 7500 transitions kmeans__k=2 75\n",
      "cp 7500 transitions kmeans__k=2 100\n",
      "cp 7500 transitions kmeans__k=2 125\n",
      "cp 7500 transitions kmeans__k=2 150\n",
      "cp 7500 transitions kmeans__k=2 200\n",
      "cp 7500 transitions kmeans__k=2 250\n",
      "sw 7500 transitions kmeans__k=2 75\n",
      "sw 7500 transitions kmeans__k=2 100\n",
      "sw 7500 transitions kmeans__k=2 125\n",
      "sw 7500 transitions kmeans__k=2 150\n",
      "sw 7500 transitions kmeans__k=2 200\n",
      "sw 7500 transitions kmeans__k=2 250\n",
      "re 7500 transitions kmeans__k=2 75\n",
      "re 7500 transitions kmeans__k=2 100\n",
      "re 7500 transitions kmeans__k=2 125\n",
      "re 7500 transitions kmeans__k=2 150\n",
      "re 7500 transitions kmeans__k=2 200\n",
      "re 7500 transitions kmeans__k=2 250\n",
      "lp 7500 transitions kmeans__k=2 75\n",
      "lp 7500 transitions kmeans__k=2 100\n",
      "lp 7500 transitions kmeans__k=2 125\n",
      "lp 7500 transitions kmeans__k=2 150\n",
      "lp 7500 transitions kmeans__k=2 200\n",
      "lp 7500 transitions kmeans__k=2 250\n",
      "cf 7500 frequency kmeans__k=2 75\n",
      "cf 7500 frequency kmeans__k=2 100\n",
      "cf 7500 frequency kmeans__k=2 125\n",
      "cf 7500 frequency kmeans__k=2 150\n",
      "cf 7500 frequency kmeans__k=2 200\n",
      "cf 7500 frequency kmeans__k=2 250\n",
      "cb 7500 frequency kmeans__k=2 75\n",
      "cb 7500 frequency kmeans__k=2 100\n",
      "cb 7500 frequency kmeans__k=2 125\n",
      "cb 7500 frequency kmeans__k=2 150\n",
      "cb 7500 frequency kmeans__k=2 200\n",
      "cb 7500 frequency kmeans__k=2 250\n",
      "cm 7500 frequency kmeans__k=2 75\n",
      "cm 7500 frequency kmeans__k=2 100\n",
      "cm 7500 frequency kmeans__k=2 125\n",
      "cm 7500 frequency kmeans__k=2 150\n",
      "cm 7500 frequency kmeans__k=2 200\n",
      "cm 7500 frequency kmeans__k=2 250\n",
      "rp 7500 frequency kmeans__k=2 75\n",
      "rp 7500 frequency kmeans__k=2 100\n",
      "rp 7500 frequency kmeans__k=2 125\n",
      "rp 7500 frequency kmeans__k=2 150\n",
      "rp 7500 frequency kmeans__k=2 200\n",
      "rp 7500 frequency kmeans__k=2 250\n",
      "pm 7500 frequency kmeans__k=2 75\n",
      "pm 7500 frequency kmeans__k=2 100\n",
      "pm 7500 frequency kmeans__k=2 125\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "pm 7500 frequency kmeans__k=2 150\n",
      "pm 7500 frequency kmeans__k=2 200\n",
      "pm 7500 frequency kmeans__k=2 250\n",
      "fr 7500 frequency kmeans__k=2 75\n",
      "fr 7500 frequency kmeans__k=2 100\n",
      "fr 7500 frequency kmeans__k=2 125\n",
      "fr 7500 frequency kmeans__k=2 150\n",
      "fr 7500 frequency kmeans__k=2 200\n",
      "fr 7500 frequency kmeans__k=2 250\n",
      "pl 7500 frequency kmeans__k=2 75\n",
      "pl 7500 frequency kmeans__k=2 100\n",
      "pl 7500 frequency kmeans__k=2 125\n",
      "pl 7500 frequency kmeans__k=2 150\n",
      "pl 7500 frequency kmeans__k=2 200\n",
      "pl 7500 frequency kmeans__k=2 250\n",
      "cd 7500 frequency kmeans__k=2 75\n",
      "cd 7500 frequency kmeans__k=2 100\n",
      "cd 7500 frequency kmeans__k=2 125\n",
      "cd 7500 frequency kmeans__k=2 150\n",
      "cd 7500 frequency kmeans__k=2 200\n",
      "cd 7500 frequency kmeans__k=2 250\n",
      "cp 7500 frequency kmeans__k=2 75\n",
      "cp 7500 frequency kmeans__k=2 100\n",
      "cp 7500 frequency kmeans__k=2 125\n",
      "cp 7500 frequency kmeans__k=2 150\n",
      "cp 7500 frequency kmeans__k=2 200\n",
      "cp 7500 frequency kmeans__k=2 250\n",
      "sw 7500 frequency kmeans__k=2 75\n",
      "sw 7500 frequency kmeans__k=2 100\n",
      "sw 7500 frequency kmeans__k=2 125\n",
      "sw 7500 frequency kmeans__k=2 150\n",
      "sw 7500 frequency kmeans__k=2 200\n",
      "sw 7500 frequency kmeans__k=2 250\n",
      "re 7500 frequency kmeans__k=2 75\n",
      "re 7500 frequency kmeans__k=2 100\n",
      "re 7500 frequency kmeans__k=2 125\n",
      "re 7500 frequency kmeans__k=2 150\n",
      "re 7500 frequency kmeans__k=2 200\n",
      "re 7500 frequency kmeans__k=2 250\n",
      "lp 7500 frequency kmeans__k=2 75\n",
      "lp 7500 frequency kmeans__k=2 100\n",
      "lp 7500 frequency kmeans__k=2 125\n",
      "lp 7500 frequency kmeans__k=2 150\n",
      "lp 7500 frequency kmeans__k=2 200\n",
      "lp 7500 frequency kmeans__k=2 250\n",
      "cf 7500 binary kmeans__k=2 75\n",
      "cf 7500 binary kmeans__k=2 100\n",
      "cf 7500 binary kmeans__k=2 125\n",
      "cf 7500 binary kmeans__k=2 150\n",
      "cf 7500 binary kmeans__k=2 200\n",
      "cf 7500 binary kmeans__k=2 250\n",
      "cb 7500 binary kmeans__k=2 75\n",
      "cb 7500 binary kmeans__k=2 100\n",
      "cb 7500 binary kmeans__k=2 125\n",
      "cb 7500 binary kmeans__k=2 150\n",
      "cb 7500 binary kmeans__k=2 200\n",
      "cb 7500 binary kmeans__k=2 250\n",
      "cm 7500 binary kmeans__k=2 75\n",
      "cm 7500 binary kmeans__k=2 100\n",
      "cm 7500 binary kmeans__k=2 125\n",
      "cm 7500 binary kmeans__k=2 150\n",
      "cm 7500 binary kmeans__k=2 200\n",
      "cm 7500 binary kmeans__k=2 250\n",
      "rp 7500 binary kmeans__k=2 75\n",
      "rp 7500 binary kmeans__k=2 100\n",
      "rp 7500 binary kmeans__k=2 125\n",
      "rp 7500 binary kmeans__k=2 150\n",
      "rp 7500 binary kmeans__k=2 200\n",
      "rp 7500 binary kmeans__k=2 250\n",
      "pm 7500 binary kmeans__k=2 75\n",
      "pm 7500 binary kmeans__k=2 100\n",
      "pm 7500 binary kmeans__k=2 125\n",
      "pm 7500 binary kmeans__k=2 150\n",
      "pm 7500 binary kmeans__k=2 200\n",
      "pm 7500 binary kmeans__k=2 250\n",
      "fr 7500 binary kmeans__k=2 75\n",
      "fr 7500 binary kmeans__k=2 100\n",
      "fr 7500 binary kmeans__k=2 125\n",
      "fr 7500 binary kmeans__k=2 150\n",
      "fr 7500 binary kmeans__k=2 200\n",
      "fr 7500 binary kmeans__k=2 250\n",
      "pl 7500 binary kmeans__k=2 75\n",
      "pl 7500 binary kmeans__k=2 100\n",
      "pl 7500 binary kmeans__k=2 125\n",
      "pl 7500 binary kmeans__k=2 150\n",
      "pl 7500 binary kmeans__k=2 200\n",
      "pl 7500 binary kmeans__k=2 250\n",
      "cd 7500 binary kmeans__k=2 75\n",
      "cd 7500 binary kmeans__k=2 100\n",
      "cd 7500 binary kmeans__k=2 125\n",
      "cd 7500 binary kmeans__k=2 150\n",
      "cd 7500 binary kmeans__k=2 200\n",
      "cd 7500 binary kmeans__k=2 250\n",
      "cp 7500 binary kmeans__k=2 75\n",
      "cp 7500 binary kmeans__k=2 100\n",
      "cp 7500 binary kmeans__k=2 125\n",
      "cp 7500 binary kmeans__k=2 150\n",
      "cp 7500 binary kmeans__k=2 200\n",
      "cp 7500 binary kmeans__k=2 250\n",
      "sw 7500 binary kmeans__k=2 75\n",
      "sw 7500 binary kmeans__k=2 100\n",
      "sw 7500 binary kmeans__k=2 125\n",
      "sw 7500 binary kmeans__k=2 150\n",
      "sw 7500 binary kmeans__k=2 200\n",
      "sw 7500 binary kmeans__k=2 250\n",
      "re 7500 binary kmeans__k=2 75\n",
      "re 7500 binary kmeans__k=2 100\n",
      "re 7500 binary kmeans__k=2 125\n",
      "re 7500 binary kmeans__k=2 150\n",
      "re 7500 binary kmeans__k=2 200\n",
      "re 7500 binary kmeans__k=2 250\n",
      "lp 7500 binary kmeans__k=2 75\n",
      "lp 7500 binary kmeans__k=2 100\n",
      "lp 7500 binary kmeans__k=2 125\n",
      "lp 7500 binary kmeans__k=2 150\n",
      "lp 7500 binary kmeans__k=2 200\n",
      "lp 7500 binary kmeans__k=2 250\n",
      "cf 7500 transitions kmeans__k=3 75\n",
      "cf 7500 transitions kmeans__k=3 100\n",
      "cf 7500 transitions kmeans__k=3 125\n",
      "cf 7500 transitions kmeans__k=3 150\n",
      "cf 7500 transitions kmeans__k=3 200\n",
      "cf 7500 transitions kmeans__k=3 250\n",
      "cb 7500 transitions kmeans__k=3 75\n",
      "cb 7500 transitions kmeans__k=3 100\n",
      "cb 7500 transitions kmeans__k=3 125\n",
      "cb 7500 transitions kmeans__k=3 150\n",
      "cb 7500 transitions kmeans__k=3 200\n",
      "cb 7500 transitions kmeans__k=3 250\n",
      "cm 7500 transitions kmeans__k=3 75\n",
      "cm 7500 transitions kmeans__k=3 100\n",
      "cm 7500 transitions kmeans__k=3 125\n",
      "cm 7500 transitions kmeans__k=3 150\n",
      "cm 7500 transitions kmeans__k=3 200\n",
      "cm 7500 transitions kmeans__k=3 250\n",
      "rp 7500 transitions kmeans__k=3 75\n",
      "rp 7500 transitions kmeans__k=3 100\n",
      "rp 7500 transitions kmeans__k=3 125\n",
      "rp 7500 transitions kmeans__k=3 150\n",
      "rp 7500 transitions kmeans__k=3 200\n",
      "rp 7500 transitions kmeans__k=3 250\n",
      "pm 7500 transitions kmeans__k=3 75\n",
      "pm 7500 transitions kmeans__k=3 100\n",
      "pm 7500 transitions kmeans__k=3 125\n",
      "pm 7500 transitions kmeans__k=3 150\n",
      "pm 7500 transitions kmeans__k=3 200\n",
      "pm 7500 transitions kmeans__k=3 250\n",
      "fr 7500 transitions kmeans__k=3 75\n",
      "fr 7500 transitions kmeans__k=3 100\n",
      "fr 7500 transitions kmeans__k=3 125\n",
      "fr 7500 transitions kmeans__k=3 150\n",
      "fr 7500 transitions kmeans__k=3 200\n",
      "fr 7500 transitions kmeans__k=3 250\n",
      "pl 7500 transitions kmeans__k=3 75\n",
      "pl 7500 transitions kmeans__k=3 100\n",
      "pl 7500 transitions kmeans__k=3 125\n",
      "pl 7500 transitions kmeans__k=3 150\n",
      "pl 7500 transitions kmeans__k=3 200\n",
      "pl 7500 transitions kmeans__k=3 250\n",
      "cd 7500 transitions kmeans__k=3 75\n",
      "cd 7500 transitions kmeans__k=3 100\n",
      "cd 7500 transitions kmeans__k=3 125\n",
      "cd 7500 transitions kmeans__k=3 150\n",
      "cd 7500 transitions kmeans__k=3 200\n",
      "cd 7500 transitions kmeans__k=3 250\n",
      "cp 7500 transitions kmeans__k=3 75\n",
      "cp 7500 transitions kmeans__k=3 100\n",
      "cp 7500 transitions kmeans__k=3 125\n",
      "cp 7500 transitions kmeans__k=3 150\n",
      "cp 7500 transitions kmeans__k=3 200\n",
      "cp 7500 transitions kmeans__k=3 250\n",
      "sw 7500 transitions kmeans__k=3 75\n",
      "sw 7500 transitions kmeans__k=3 100\n",
      "sw 7500 transitions kmeans__k=3 125\n",
      "sw 7500 transitions kmeans__k=3 150\n",
      "sw 7500 transitions kmeans__k=3 200\n",
      "sw 7500 transitions kmeans__k=3 250\n",
      "re 7500 transitions kmeans__k=3 75\n",
      "re 7500 transitions kmeans__k=3 100\n",
      "re 7500 transitions kmeans__k=3 125\n",
      "re 7500 transitions kmeans__k=3 150\n",
      "re 7500 transitions kmeans__k=3 200\n",
      "re 7500 transitions kmeans__k=3 250\n",
      "lp 7500 transitions kmeans__k=3 75\n",
      "lp 7500 transitions kmeans__k=3 100\n",
      "lp 7500 transitions kmeans__k=3 125\n",
      "lp 7500 transitions kmeans__k=3 150\n",
      "lp 7500 transitions kmeans__k=3 200\n",
      "lp 7500 transitions kmeans__k=3 250\n",
      "cf 7500 frequency kmeans__k=3 75\n",
      "cf 7500 frequency kmeans__k=3 100\n",
      "cf 7500 frequency kmeans__k=3 125\n",
      "cf 7500 frequency kmeans__k=3 150\n",
      "cf 7500 frequency kmeans__k=3 200\n",
      "cf 7500 frequency kmeans__k=3 250\n",
      "cb 7500 frequency kmeans__k=3 75\n",
      "cb 7500 frequency kmeans__k=3 100\n",
      "cb 7500 frequency kmeans__k=3 125\n",
      "cb 7500 frequency kmeans__k=3 150\n",
      "cb 7500 frequency kmeans__k=3 200\n",
      "cb 7500 frequency kmeans__k=3 250\n",
      "cm 7500 frequency kmeans__k=3 75\n",
      "cm 7500 frequency kmeans__k=3 100\n",
      "cm 7500 frequency kmeans__k=3 125\n",
      "cm 7500 frequency kmeans__k=3 150\n",
      "cm 7500 frequency kmeans__k=3 200\n",
      "cm 7500 frequency kmeans__k=3 250\n",
      "rp 7500 frequency kmeans__k=3 75\n",
      "rp 7500 frequency kmeans__k=3 100\n",
      "rp 7500 frequency kmeans__k=3 125\n",
      "rp 7500 frequency kmeans__k=3 150\n",
      "rp 7500 frequency kmeans__k=3 200\n",
      "rp 7500 frequency kmeans__k=3 250\n",
      "pm 7500 frequency kmeans__k=3 75\n",
      "pm 7500 frequency kmeans__k=3 100\n",
      "pm 7500 frequency kmeans__k=3 125\n",
      "pm 7500 frequency kmeans__k=3 150\n",
      "pm 7500 frequency kmeans__k=3 200\n",
      "pm 7500 frequency kmeans__k=3 250\n",
      "fr 7500 frequency kmeans__k=3 75\n",
      "fr 7500 frequency kmeans__k=3 100\n",
      "fr 7500 frequency kmeans__k=3 125\n",
      "fr 7500 frequency kmeans__k=3 150\n",
      "fr 7500 frequency kmeans__k=3 200\n",
      "fr 7500 frequency kmeans__k=3 250\n",
      "pl 7500 frequency kmeans__k=3 75\n",
      "pl 7500 frequency kmeans__k=3 100\n",
      "pl 7500 frequency kmeans__k=3 125\n",
      "pl 7500 frequency kmeans__k=3 150\n",
      "pl 7500 frequency kmeans__k=3 200\n",
      "pl 7500 frequency kmeans__k=3 250\n",
      "cd 7500 frequency kmeans__k=3 75\n",
      "cd 7500 frequency kmeans__k=3 100\n",
      "cd 7500 frequency kmeans__k=3 125\n",
      "cd 7500 frequency kmeans__k=3 150\n",
      "cd 7500 frequency kmeans__k=3 200\n",
      "cd 7500 frequency kmeans__k=3 250\n",
      "cp 7500 frequency kmeans__k=3 75\n",
      "cp 7500 frequency kmeans__k=3 100\n",
      "cp 7500 frequency kmeans__k=3 125\n",
      "cp 7500 frequency kmeans__k=3 150\n",
      "cp 7500 frequency kmeans__k=3 200\n",
      "cp 7500 frequency kmeans__k=3 250\n",
      "sw 7500 frequency kmeans__k=3 75\n",
      "sw 7500 frequency kmeans__k=3 100\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "sw 7500 frequency kmeans__k=3 125\n",
      "sw 7500 frequency kmeans__k=3 150\n",
      "sw 7500 frequency kmeans__k=3 200\n",
      "sw 7500 frequency kmeans__k=3 250\n",
      "re 7500 frequency kmeans__k=3 75\n",
      "re 7500 frequency kmeans__k=3 100\n",
      "re 7500 frequency kmeans__k=3 125\n",
      "re 7500 frequency kmeans__k=3 150\n",
      "re 7500 frequency kmeans__k=3 200\n",
      "re 7500 frequency kmeans__k=3 250\n",
      "lp 7500 frequency kmeans__k=3 75\n",
      "lp 7500 frequency kmeans__k=3 100\n",
      "lp 7500 frequency kmeans__k=3 125\n",
      "lp 7500 frequency kmeans__k=3 150\n",
      "lp 7500 frequency kmeans__k=3 200\n",
      "lp 7500 frequency kmeans__k=3 250\n",
      "cf 7500 binary kmeans__k=3 75\n",
      "cf 7500 binary kmeans__k=3 100\n",
      "cf 7500 binary kmeans__k=3 125\n",
      "cf 7500 binary kmeans__k=3 150\n",
      "cf 7500 binary kmeans__k=3 200\n",
      "cf 7500 binary kmeans__k=3 250\n",
      "cb 7500 binary kmeans__k=3 75\n",
      "cb 7500 binary kmeans__k=3 100\n",
      "cb 7500 binary kmeans__k=3 125\n",
      "cb 7500 binary kmeans__k=3 150\n",
      "cb 7500 binary kmeans__k=3 200\n",
      "cb 7500 binary kmeans__k=3 250\n",
      "cm 7500 binary kmeans__k=3 75\n",
      "cm 7500 binary kmeans__k=3 100\n",
      "cm 7500 binary kmeans__k=3 125\n",
      "cm 7500 binary kmeans__k=3 150\n",
      "cm 7500 binary kmeans__k=3 200\n",
      "cm 7500 binary kmeans__k=3 250\n",
      "rp 7500 binary kmeans__k=3 75\n",
      "rp 7500 binary kmeans__k=3 100\n",
      "rp 7500 binary kmeans__k=3 125\n",
      "rp 7500 binary kmeans__k=3 150\n",
      "rp 7500 binary kmeans__k=3 200\n",
      "rp 7500 binary kmeans__k=3 250\n",
      "pm 7500 binary kmeans__k=3 75\n",
      "pm 7500 binary kmeans__k=3 100\n",
      "pm 7500 binary kmeans__k=3 125\n",
      "pm 7500 binary kmeans__k=3 150\n",
      "pm 7500 binary kmeans__k=3 200\n",
      "pm 7500 binary kmeans__k=3 250\n",
      "fr 7500 binary kmeans__k=3 75\n",
      "fr 7500 binary kmeans__k=3 100\n",
      "fr 7500 binary kmeans__k=3 125\n",
      "fr 7500 binary kmeans__k=3 150\n",
      "fr 7500 binary kmeans__k=3 200\n",
      "fr 7500 binary kmeans__k=3 250\n",
      "pl 7500 binary kmeans__k=3 75\n",
      "pl 7500 binary kmeans__k=3 100\n",
      "pl 7500 binary kmeans__k=3 125\n",
      "pl 7500 binary kmeans__k=3 150\n",
      "pl 7500 binary kmeans__k=3 200\n",
      "pl 7500 binary kmeans__k=3 250\n",
      "cd 7500 binary kmeans__k=3 75\n",
      "cd 7500 binary kmeans__k=3 100\n",
      "cd 7500 binary kmeans__k=3 125\n",
      "cd 7500 binary kmeans__k=3 150\n",
      "cd 7500 binary kmeans__k=3 200\n",
      "cd 7500 binary kmeans__k=3 250\n",
      "cp 7500 binary kmeans__k=3 75\n",
      "cp 7500 binary kmeans__k=3 100\n",
      "cp 7500 binary kmeans__k=3 125\n",
      "cp 7500 binary kmeans__k=3 150\n",
      "cp 7500 binary kmeans__k=3 200\n",
      "cp 7500 binary kmeans__k=3 250\n",
      "sw 7500 binary kmeans__k=3 75\n",
      "sw 7500 binary kmeans__k=3 100\n",
      "sw 7500 binary kmeans__k=3 125\n",
      "sw 7500 binary kmeans__k=3 150\n",
      "sw 7500 binary kmeans__k=3 200\n",
      "sw 7500 binary kmeans__k=3 250\n",
      "re 7500 binary kmeans__k=3 75\n",
      "re 7500 binary kmeans__k=3 100\n",
      "re 7500 binary kmeans__k=3 125\n",
      "re 7500 binary kmeans__k=3 150\n",
      "re 7500 binary kmeans__k=3 200\n",
      "re 7500 binary kmeans__k=3 250\n",
      "lp 7500 binary kmeans__k=3 75\n",
      "lp 7500 binary kmeans__k=3 100\n",
      "lp 7500 binary kmeans__k=3 125\n",
      "lp 7500 binary kmeans__k=3 150\n",
      "lp 7500 binary kmeans__k=3 200\n",
      "lp 7500 binary kmeans__k=3 250\n"
     ]
    }
   ],
   "source": [
    "resp__2 = []\n",
    "\n",
    "for c in combinations:\n",
    "    col_names, y_true = cached_resp[c['tipo_mudanca'] + \"_\" + str(c['log_size']) + \"_\" + c['representation']]\n",
    "    \n",
    "    r = {}\n",
    "    r.update(c)\n",
    "    \n",
    "    r[\"y_true\"] = y_true\n",
    "    \n",
    "    print(c[\"tipo_mudanca\"], c[\"log_size\"], c[\"representation\"], c[\"model\"], c[\"window_size\"])\n",
    "    \n",
    "    df = pd.read_csv(base_dir + c[\"tipo_mudanca\"] + \"/\" + opts_to_string(c) + \".csv\", index_col=0)\n",
    "    try:\n",
    "        df.centroids = df.centroids.apply(string_to_numpy)\n",
    "    except:\n",
    "        print(\"ERROR\")\n",
    "        continue\n",
    "    df.index += df.index[1]\n",
    "    \n",
    "    drifts = []\n",
    "    roc_scores = []\n",
    "    avg_precisions = []\n",
    "    \n",
    "    previous_window = 0\n",
    "    step = int(c['log_size']/10)\n",
    "    for i in range(step, c['log_size'], step): \n",
    "        # print(i, df.index[previous_window], df.index[df.index.get_loc(i, method=\"nearest\")], df.index[df.index.get_loc(i + c['window_size'], method=\"nearest\")])\n",
    "        # i-(2*c['window_size']):\n",
    "        \n",
    "        local = localize_drift(\n",
    "            df.centroids.iloc[previous_window:df.index.get_loc(i, method=\"nearest\")], \n",
    "            df.centroids.iloc[df.index.get_loc(i + c['window_size'], method=\"nearest\")],\n",
    "            col_names=col_names\n",
    "        )\n",
    "        \n",
    "        previous_window = df.index.get_loc(i + c['window_size'], method=\"nearest\") # + 1\n",
    "        \n",
    "        y_score = local.values.reshape(1,-1)[0]\n",
    "        \n",
    "        roc_scores.append(\n",
    "            roc_auc_score(r[\"y_true\"], y_score)\n",
    "        )\n",
    "        \n",
    "        avg_precisions.append(\n",
    "            average_precision_score(r[\"y_true\"], y_score)\n",
    "        )\n",
    "        \n",
    "        drifts.append(local.sort_values(0, ascending=False)[local[0] > 0].to_dict()[0])\n",
    "    \n",
    "    r[\"localizations\"] = drifts\n",
    "    r[\"roc_scores\"] = roc_scores\n",
    "    r[\"avg_precisions\"] = avg_precisions\n",
    "    \n",
    "    r[\"AVG_roc_scores\"] = np.mean(roc_scores)\n",
    "    r[\"AVG_avg_precisions\"] = np.mean(avg_precisions)\n",
    "    resp__2.append(r)    \n",
    "    \n",
    "#    break"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>log_size</th>\n",
       "      <th>model</th>\n",
       "      <th>representation</th>\n",
       "      <th>sliding_window</th>\n",
       "      <th>tipo_mudanca</th>\n",
       "      <th>window_size</th>\n",
       "      <th>y_true</th>\n",
       "      <th>localizations</th>\n",
       "      <th>roc_scores</th>\n",
       "      <th>avg_precisions</th>\n",
       "      <th>AVG_roc_scores</th>\n",
       "      <th>AVG_avg_precisions</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>10000</td>\n",
       "      <td>kmeans__k=2</td>\n",
       "      <td>transitions</td>\n",
       "      <td>False</td>\n",
       "      <td>cf</td>\n",
       "      <td>75</td>\n",
       "      <td>[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, ...</td>\n",
       "      <td>[{'M N': 0.22222222444444445, 'M O': 0.0665006...</td>\n",
       "      <td>[0.9772727272727272, 0.9886363636363636, 0.988...</td>\n",
       "      <td>[0.8875, 0.95, 0.95, 0.95, 1.0, 0.804166666666...</td>\n",
       "      <td>0.981692</td>\n",
       "      <td>0.894444</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>10000</td>\n",
       "      <td>kmeans__k=2</td>\n",
       "      <td>transitions</td>\n",
       "      <td>False</td>\n",
       "      <td>cf</td>\n",
       "      <td>100</td>\n",
       "      <td>[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, ...</td>\n",
       "      <td>[{'M N': 0.5, 'M O': 0.1408642408780667, 'N O'...</td>\n",
       "      <td>[0.9715909090909091, 0.9886363636363636, 0.988...</td>\n",
       "      <td>[0.8166666666666667, 0.95, 0.95, 0.85000000000...</td>\n",
       "      <td>0.980429</td>\n",
       "      <td>0.867130</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>10000</td>\n",
       "      <td>kmeans__k=2</td>\n",
       "      <td>transitions</td>\n",
       "      <td>False</td>\n",
       "      <td>cf</td>\n",
       "      <td>125</td>\n",
       "      <td>[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, ...</td>\n",
       "      <td>[{'M N': 0.5, 'N O': 0.13857436956133617, 'M O...</td>\n",
       "      <td>[0.9715909090909091, 0.9943181818181818, 0.988...</td>\n",
       "      <td>[0.8041666666666667, 0.95, 0.95, 0.95, 1.0, 0....</td>\n",
       "      <td>0.981061</td>\n",
       "      <td>0.877315</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>10000</td>\n",
       "      <td>kmeans__k=2</td>\n",
       "      <td>transitions</td>\n",
       "      <td>False</td>\n",
       "      <td>cf</td>\n",
       "      <td>150</td>\n",
       "      <td>[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, ...</td>\n",
       "      <td>[{'M N': 0.5, 'N O': 0.140847231039704, 'M O':...</td>\n",
       "      <td>[0.9715909090909091, 0.9943181818181818, 0.988...</td>\n",
       "      <td>[0.8041666666666667, 0.95, 0.95, 0.95, 0.95, 0...</td>\n",
       "      <td>0.982323</td>\n",
       "      <td>0.898148</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>10000</td>\n",
       "      <td>kmeans__k=2</td>\n",
       "      <td>transitions</td>\n",
       "      <td>False</td>\n",
       "      <td>cf</td>\n",
       "      <td>200</td>\n",
       "      <td>[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, ...</td>\n",
       "      <td>[{'M N': 0.5, 'N O': 0.13391271574047678, 'M O...</td>\n",
       "      <td>[0.9715909090909091, 1.0, 0.9886363636363636, ...</td>\n",
       "      <td>[0.8166666666666667, 1.0, 0.95, 0.95, 0.95, 0....</td>\n",
       "      <td>0.981061</td>\n",
       "      <td>0.880093</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1291</th>\n",
       "      <td>7500</td>\n",
       "      <td>kmeans__k=3</td>\n",
       "      <td>binary</td>\n",
       "      <td>False</td>\n",
       "      <td>lp</td>\n",
       "      <td>100</td>\n",
       "      <td>[1, 0, 1, 1, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, ...</td>\n",
       "      <td>[{'E': 0.12244897959183675, 'END_A': 0.1224489...</td>\n",
       "      <td>[0.23333333333333334, 0.23333333333333334, 0.2...</td>\n",
       "      <td>[0.21052631578947367, 0.21052631578947367, 0.2...</td>\n",
       "      <td>0.233333</td>\n",
       "      <td>0.210526</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1292</th>\n",
       "      <td>7500</td>\n",
       "      <td>kmeans__k=3</td>\n",
       "      <td>binary</td>\n",
       "      <td>False</td>\n",
       "      <td>lp</td>\n",
       "      <td>125</td>\n",
       "      <td>[1, 0, 1, 1, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, ...</td>\n",
       "      <td>[{'N': 0.036978619968428945, 'M': 0.0369786199...</td>\n",
       "      <td>[0.23333333333333334, 0.23333333333333334, 0.2...</td>\n",
       "      <td>[0.21052631578947367, 0.21052631578947367, 0.2...</td>\n",
       "      <td>0.233333</td>\n",
       "      <td>0.210526</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1293</th>\n",
       "      <td>7500</td>\n",
       "      <td>kmeans__k=3</td>\n",
       "      <td>binary</td>\n",
       "      <td>False</td>\n",
       "      <td>lp</td>\n",
       "      <td>150</td>\n",
       "      <td>[1, 0, 1, 1, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, ...</td>\n",
       "      <td>[{'E': 0.16666666666666666, 'END_A': 0.1666666...</td>\n",
       "      <td>[0.23333333333333334, 0.23333333333333334, 0.2...</td>\n",
       "      <td>[0.21052631578947367, 0.21052631578947367, 0.2...</td>\n",
       "      <td>0.233333</td>\n",
       "      <td>0.210526</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1294</th>\n",
       "      <td>7500</td>\n",
       "      <td>kmeans__k=3</td>\n",
       "      <td>binary</td>\n",
       "      <td>False</td>\n",
       "      <td>lp</td>\n",
       "      <td>200</td>\n",
       "      <td>[1, 0, 1, 1, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, ...</td>\n",
       "      <td>[{'E': 0.07407407407407408, 'END_A': 0.0740740...</td>\n",
       "      <td>[0.23333333333333334, 0.23333333333333334, 0.3...</td>\n",
       "      <td>[0.21052631578947367, 0.21052631578947367, 0.2...</td>\n",
       "      <td>0.248148</td>\n",
       "      <td>0.210526</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1295</th>\n",
       "      <td>7500</td>\n",
       "      <td>kmeans__k=3</td>\n",
       "      <td>binary</td>\n",
       "      <td>False</td>\n",
       "      <td>lp</td>\n",
       "      <td>250</td>\n",
       "      <td>[1, 0, 1, 1, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, ...</td>\n",
       "      <td>[{'E': 0.16666666666666666, 'END_A': 0.1666666...</td>\n",
       "      <td>[0.23333333333333334, 0.3666666666666667, 0.23...</td>\n",
       "      <td>[0.21052631578947367, 0.21052631578947367, 0.2...</td>\n",
       "      <td>0.277778</td>\n",
       "      <td>0.210526</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>1296 rows × 12 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "      log_size        model representation  sliding_window tipo_mudanca  \\\n",
       "0        10000  kmeans__k=2    transitions           False           cf   \n",
       "1        10000  kmeans__k=2    transitions           False           cf   \n",
       "2        10000  kmeans__k=2    transitions           False           cf   \n",
       "3        10000  kmeans__k=2    transitions           False           cf   \n",
       "4        10000  kmeans__k=2    transitions           False           cf   \n",
       "...        ...          ...            ...             ...          ...   \n",
       "1291      7500  kmeans__k=3         binary           False           lp   \n",
       "1292      7500  kmeans__k=3         binary           False           lp   \n",
       "1293      7500  kmeans__k=3         binary           False           lp   \n",
       "1294      7500  kmeans__k=3         binary           False           lp   \n",
       "1295      7500  kmeans__k=3         binary           False           lp   \n",
       "\n",
       "      window_size                                             y_true  \\\n",
       "0              75  [0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, ...   \n",
       "1             100  [0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, ...   \n",
       "2             125  [0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, ...   \n",
       "3             150  [0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, ...   \n",
       "4             200  [0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, ...   \n",
       "...           ...                                                ...   \n",
       "1291          100  [1, 0, 1, 1, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, ...   \n",
       "1292          125  [1, 0, 1, 1, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, ...   \n",
       "1293          150  [1, 0, 1, 1, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, ...   \n",
       "1294          200  [1, 0, 1, 1, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, ...   \n",
       "1295          250  [1, 0, 1, 1, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, ...   \n",
       "\n",
       "                                          localizations  \\\n",
       "0     [{'M N': 0.22222222444444445, 'M O': 0.0665006...   \n",
       "1     [{'M N': 0.5, 'M O': 0.1408642408780667, 'N O'...   \n",
       "2     [{'M N': 0.5, 'N O': 0.13857436956133617, 'M O...   \n",
       "3     [{'M N': 0.5, 'N O': 0.140847231039704, 'M O':...   \n",
       "4     [{'M N': 0.5, 'N O': 0.13391271574047678, 'M O...   \n",
       "...                                                 ...   \n",
       "1291  [{'E': 0.12244897959183675, 'END_A': 0.1224489...   \n",
       "1292  [{'N': 0.036978619968428945, 'M': 0.0369786199...   \n",
       "1293  [{'E': 0.16666666666666666, 'END_A': 0.1666666...   \n",
       "1294  [{'E': 0.07407407407407408, 'END_A': 0.0740740...   \n",
       "1295  [{'E': 0.16666666666666666, 'END_A': 0.1666666...   \n",
       "\n",
       "                                             roc_scores  \\\n",
       "0     [0.9772727272727272, 0.9886363636363636, 0.988...   \n",
       "1     [0.9715909090909091, 0.9886363636363636, 0.988...   \n",
       "2     [0.9715909090909091, 0.9943181818181818, 0.988...   \n",
       "3     [0.9715909090909091, 0.9943181818181818, 0.988...   \n",
       "4     [0.9715909090909091, 1.0, 0.9886363636363636, ...   \n",
       "...                                                 ...   \n",
       "1291  [0.23333333333333334, 0.23333333333333334, 0.2...   \n",
       "1292  [0.23333333333333334, 0.23333333333333334, 0.2...   \n",
       "1293  [0.23333333333333334, 0.23333333333333334, 0.2...   \n",
       "1294  [0.23333333333333334, 0.23333333333333334, 0.3...   \n",
       "1295  [0.23333333333333334, 0.3666666666666667, 0.23...   \n",
       "\n",
       "                                         avg_precisions  AVG_roc_scores  \\\n",
       "0     [0.8875, 0.95, 0.95, 0.95, 1.0, 0.804166666666...        0.981692   \n",
       "1     [0.8166666666666667, 0.95, 0.95, 0.85000000000...        0.980429   \n",
       "2     [0.8041666666666667, 0.95, 0.95, 0.95, 1.0, 0....        0.981061   \n",
       "3     [0.8041666666666667, 0.95, 0.95, 0.95, 0.95, 0...        0.982323   \n",
       "4     [0.8166666666666667, 1.0, 0.95, 0.95, 0.95, 0....        0.981061   \n",
       "...                                                 ...             ...   \n",
       "1291  [0.21052631578947367, 0.21052631578947367, 0.2...        0.233333   \n",
       "1292  [0.21052631578947367, 0.21052631578947367, 0.2...        0.233333   \n",
       "1293  [0.21052631578947367, 0.21052631578947367, 0.2...        0.233333   \n",
       "1294  [0.21052631578947367, 0.21052631578947367, 0.2...        0.248148   \n",
       "1295  [0.21052631578947367, 0.21052631578947367, 0.2...        0.277778   \n",
       "\n",
       "      AVG_avg_precisions  \n",
       "0               0.894444  \n",
       "1               0.867130  \n",
       "2               0.877315  \n",
       "3               0.898148  \n",
       "4               0.880093  \n",
       "...                  ...  \n",
       "1291            0.210526  \n",
       "1292            0.210526  \n",
       "1293            0.210526  \n",
       "1294            0.210526  \n",
       "1295            0.210526  \n",
       "\n",
       "[1296 rows x 12 columns]"
      ]
     },
     "execution_count": 35,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "resp_df = pd.DataFrame.from_dict(resp__2)\n",
    "resp_df"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [],
   "source": [
    "resp_df[\"flag_binary_support\"] = resp_df[\"tipo_mudanca\"].isin([x for x in binary_support if x != 'fr'] + ['re'])\n",
    "resp_df[\"flag_frequency_support\"] = resp_df[\"tipo_mudanca\"].isin([x for x in binary_support if x != 'fr'] + ['re', 'lp', 'cp'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 532,
   "metadata": {},
   "outputs": [],
   "source": [
    "resp_df.to_excel(\"results_localization_experiments_v3.xlsx\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 287,
   "metadata": {},
   "outputs": [],
   "source": [
    "resp_df.to_pickle(\"results_localization_experiments.pickle\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 156,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Index(['log_size', 'model', 'representation', 'sliding_window', 'tipo_mudanca',\n",
       "       'window_size', 'y_true', 'localizations', 'roc_scores',\n",
       "       'avg_precisions', 'AVG_roc_scores', 'AVG_avg_precisions',\n",
       "       'flag_binary_support', 'flag_frequency_support'],\n",
       "      dtype='object')"
      ]
     },
     "execution_count": 156,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "resp_df.columns"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>model</th>\n",
       "      <th>representation</th>\n",
       "      <th>sliding_window</th>\n",
       "      <th>tipo_mudanca</th>\n",
       "      <th>window_size</th>\n",
       "      <th>AVG_roc_scores</th>\n",
       "      <th>AVG_avg_precisions</th>\n",
       "      <th>flag_binary_support</th>\n",
       "      <th>flag_frequency_support</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>kmeans__k=2</td>\n",
       "      <td>binary</td>\n",
       "      <td>False</td>\n",
       "      <td>cb</td>\n",
       "      <td>75</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td>kmeans__k=2</td>\n",
       "      <td>binary</td>\n",
       "      <td>False</td>\n",
       "      <td>cf</td>\n",
       "      <td>75</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>30</th>\n",
       "      <td>kmeans__k=2</td>\n",
       "      <td>binary</td>\n",
       "      <td>False</td>\n",
       "      <td>fr</td>\n",
       "      <td>75</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>48</th>\n",
       "      <td>kmeans__k=2</td>\n",
       "      <td>binary</td>\n",
       "      <td>False</td>\n",
       "      <td>pm</td>\n",
       "      <td>75</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>54</th>\n",
       "      <td>kmeans__k=2</td>\n",
       "      <td>binary</td>\n",
       "      <td>False</td>\n",
       "      <td>re</td>\n",
       "      <td>75</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>60</th>\n",
       "      <td>kmeans__k=2</td>\n",
       "      <td>binary</td>\n",
       "      <td>False</td>\n",
       "      <td>rp</td>\n",
       "      <td>75</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>96</th>\n",
       "      <td>kmeans__k=2</td>\n",
       "      <td>frequency</td>\n",
       "      <td>False</td>\n",
       "      <td>cp</td>\n",
       "      <td>75</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>326</th>\n",
       "      <td>kmeans__k=3</td>\n",
       "      <td>frequency</td>\n",
       "      <td>False</td>\n",
       "      <td>lp</td>\n",
       "      <td>125</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>164</th>\n",
       "      <td>kmeans__k=2</td>\n",
       "      <td>transitions</td>\n",
       "      <td>False</td>\n",
       "      <td>cm</td>\n",
       "      <td>125</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>211</th>\n",
       "      <td>kmeans__k=2</td>\n",
       "      <td>transitions</td>\n",
       "      <td>False</td>\n",
       "      <td>sw</td>\n",
       "      <td>100</td>\n",
       "      <td>0.927273</td>\n",
       "      <td>0.829419</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>191</th>\n",
       "      <td>kmeans__k=2</td>\n",
       "      <td>transitions</td>\n",
       "      <td>False</td>\n",
       "      <td>pl</td>\n",
       "      <td>250</td>\n",
       "      <td>0.922593</td>\n",
       "      <td>0.669042</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>155</th>\n",
       "      <td>kmeans__k=2</td>\n",
       "      <td>transitions</td>\n",
       "      <td>False</td>\n",
       "      <td>cd</td>\n",
       "      <td>250</td>\n",
       "      <td>0.825930</td>\n",
       "      <td>0.287863</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "           model representation  sliding_window tipo_mudanca  window_size  \\\n",
       "0    kmeans__k=2         binary           False           cb           75   \n",
       "12   kmeans__k=2         binary           False           cf           75   \n",
       "30   kmeans__k=2         binary           False           fr           75   \n",
       "48   kmeans__k=2         binary           False           pm           75   \n",
       "54   kmeans__k=2         binary           False           re           75   \n",
       "60   kmeans__k=2         binary           False           rp           75   \n",
       "96   kmeans__k=2      frequency           False           cp           75   \n",
       "326  kmeans__k=3      frequency           False           lp          125   \n",
       "164  kmeans__k=2    transitions           False           cm          125   \n",
       "211  kmeans__k=2    transitions           False           sw          100   \n",
       "191  kmeans__k=2    transitions           False           pl          250   \n",
       "155  kmeans__k=2    transitions           False           cd          250   \n",
       "\n",
       "     AVG_roc_scores  AVG_avg_precisions  flag_binary_support  \\\n",
       "0          1.000000            1.000000                 True   \n",
       "12         1.000000            1.000000                 True   \n",
       "30         1.000000            1.000000                False   \n",
       "48         1.000000            1.000000                 True   \n",
       "54         1.000000            1.000000                 True   \n",
       "60         1.000000            1.000000                 True   \n",
       "96         1.000000            1.000000                False   \n",
       "326        1.000000            1.000000                False   \n",
       "164        1.000000            1.000000                 True   \n",
       "211        0.927273            0.829419                False   \n",
       "191        0.922593            0.669042                False   \n",
       "155        0.825930            0.287863                False   \n",
       "\n",
       "     flag_frequency_support  \n",
       "0                      True  \n",
       "12                     True  \n",
       "30                    False  \n",
       "48                     True  \n",
       "54                     True  \n",
       "60                     True  \n",
       "96                     True  \n",
       "326                    True  \n",
       "164                    True  \n",
       "211                   False  \n",
       "191                   False  \n",
       "155                   False  "
      ]
     },
     "execution_count": 37,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "resp_df_grouped = resp_df.groupby([\n",
    "    'model', 'representation', 'sliding_window', 'tipo_mudanca', 'window_size',\n",
    "]).agg({\n",
    "    'AVG_roc_scores': 'mean',\n",
    "    'AVG_avg_precisions': 'mean',\n",
    "    'flag_binary_support': 'max',\n",
    "    'flag_frequency_support': 'max',\n",
    "}).reset_index()\n",
    "\n",
    "resp_df_grouped.sort_values([\"AVG_roc_scores\", \"window_size\", 'representation', \"model\"],ascending=[False, True, True, True]).drop_duplicates(\"tipo_mudanca\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>model</th>\n",
       "      <th>representation</th>\n",
       "      <th>sliding_window</th>\n",
       "      <th>window_size</th>\n",
       "      <th>AVG_roc_scores</th>\n",
       "      <th>AVG_avg_precisions</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>17</th>\n",
       "      <td>kmeans__k=2</td>\n",
       "      <td>transitions</td>\n",
       "      <td>False</td>\n",
       "      <td>250</td>\n",
       "      <td>0.921613</td>\n",
       "      <td>0.744257</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13</th>\n",
       "      <td>kmeans__k=2</td>\n",
       "      <td>transitions</td>\n",
       "      <td>False</td>\n",
       "      <td>100</td>\n",
       "      <td>0.918120</td>\n",
       "      <td>0.739732</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14</th>\n",
       "      <td>kmeans__k=2</td>\n",
       "      <td>transitions</td>\n",
       "      <td>False</td>\n",
       "      <td>125</td>\n",
       "      <td>0.916649</td>\n",
       "      <td>0.740208</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td>kmeans__k=2</td>\n",
       "      <td>transitions</td>\n",
       "      <td>False</td>\n",
       "      <td>75</td>\n",
       "      <td>0.909638</td>\n",
       "      <td>0.735168</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>16</th>\n",
       "      <td>kmeans__k=2</td>\n",
       "      <td>transitions</td>\n",
       "      <td>False</td>\n",
       "      <td>200</td>\n",
       "      <td>0.909015</td>\n",
       "      <td>0.733327</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "          model representation  sliding_window  window_size  AVG_roc_scores  \\\n",
       "17  kmeans__k=2    transitions           False          250        0.921613   \n",
       "13  kmeans__k=2    transitions           False          100        0.918120   \n",
       "14  kmeans__k=2    transitions           False          125        0.916649   \n",
       "12  kmeans__k=2    transitions           False           75        0.909638   \n",
       "16  kmeans__k=2    transitions           False          200        0.909015   \n",
       "\n",
       "    AVG_avg_precisions  \n",
       "17            0.744257  \n",
       "13            0.739732  \n",
       "14            0.740208  \n",
       "12            0.735168  \n",
       "16            0.733327  "
      ]
     },
     "execution_count": 38,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "resp_df_grouped.groupby(['model', 'representation', 'sliding_window', 'window_size']).agg({\n",
    "    'AVG_roc_scores': 'mean',\n",
    "    'AVG_avg_precisions': 'mean',\n",
    "}).reset_index().sort_values([\"AVG_roc_scores\", \"window_size\", 'representation', \"model\"],ascending=[False, True, True, True]).head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 174,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>log_size</th>\n",
       "      <th>model</th>\n",
       "      <th>representation</th>\n",
       "      <th>sliding_window</th>\n",
       "      <th>tipo_mudanca</th>\n",
       "      <th>window_size</th>\n",
       "      <th>y_true</th>\n",
       "      <th>localizations</th>\n",
       "      <th>roc_scores</th>\n",
       "      <th>avg_precisions</th>\n",
       "      <th>AVG_roc_scores</th>\n",
       "      <th>AVG_avg_precisions</th>\n",
       "      <th>flag_binary_support</th>\n",
       "      <th>flag_frequency_support</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>86</th>\n",
       "      <td>10000</td>\n",
       "      <td>kmeans__k=2</td>\n",
       "      <td>frequency</td>\n",
       "      <td>False</td>\n",
       "      <td>cm</td>\n",
       "      <td>125</td>\n",
       "      <td>[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, ...</td>\n",
       "      <td>[{'I': 0.278081009813272, 'M': 0.0145380202341...</td>\n",
       "      <td>[1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0]</td>\n",
       "      <td>[1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0]</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.0</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>518</th>\n",
       "      <td>5000</td>\n",
       "      <td>kmeans__k=2</td>\n",
       "      <td>frequency</td>\n",
       "      <td>False</td>\n",
       "      <td>cm</td>\n",
       "      <td>125</td>\n",
       "      <td>[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, ...</td>\n",
       "      <td>[{'I': 0.25215291080235247, 'N': 0.01081131648...</td>\n",
       "      <td>[1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0]</td>\n",
       "      <td>[1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0]</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.0</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>950</th>\n",
       "      <td>7500</td>\n",
       "      <td>kmeans__k=2</td>\n",
       "      <td>frequency</td>\n",
       "      <td>False</td>\n",
       "      <td>cm</td>\n",
       "      <td>125</td>\n",
       "      <td>[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, ...</td>\n",
       "      <td>[{'H': 0.12093523310937918, 'I': 0.12093523310...</td>\n",
       "      <td>[0.9722222222222222, 0.9722222222222222, 0.972...</td>\n",
       "      <td>[0.5, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5]</td>\n",
       "      <td>0.972222</td>\n",
       "      <td>0.5</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "     log_size        model representation  sliding_window tipo_mudanca  \\\n",
       "86      10000  kmeans__k=2      frequency           False           cm   \n",
       "518      5000  kmeans__k=2      frequency           False           cm   \n",
       "950      7500  kmeans__k=2      frequency           False           cm   \n",
       "\n",
       "     window_size                                             y_true  \\\n",
       "86           125  [0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, ...   \n",
       "518          125  [0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, ...   \n",
       "950          125  [0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, ...   \n",
       "\n",
       "                                         localizations  \\\n",
       "86   [{'I': 0.278081009813272, 'M': 0.0145380202341...   \n",
       "518  [{'I': 0.25215291080235247, 'N': 0.01081131648...   \n",
       "950  [{'H': 0.12093523310937918, 'I': 0.12093523310...   \n",
       "\n",
       "                                            roc_scores  \\\n",
       "86       [1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0]   \n",
       "518      [1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0]   \n",
       "950  [0.9722222222222222, 0.9722222222222222, 0.972...   \n",
       "\n",
       "                                    avg_precisions  AVG_roc_scores  \\\n",
       "86   [1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0]        1.000000   \n",
       "518  [1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0]        1.000000   \n",
       "950  [0.5, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5]        0.972222   \n",
       "\n",
       "     AVG_avg_precisions  flag_binary_support  flag_frequency_support  \n",
       "86                  1.0                 True                    True  \n",
       "518                 1.0                 True                    True  \n",
       "950                 0.5                 True                    True  "
      ]
     },
     "execution_count": 174,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "resp_df[\n",
    "    (resp_df.model == \"kmeans__k=2\")\n",
    "    & (resp_df.representation == \"frequency\")\n",
    "    & (resp_df.window_size == 125)\n",
    "    & (resp_df.tipo_mudanca == \"cm\")\n",
    "]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>model</th>\n",
       "      <th>representation</th>\n",
       "      <th>sliding_window</th>\n",
       "      <th>tipo_mudanca</th>\n",
       "      <th>window_size</th>\n",
       "      <th>AVG_roc_scores</th>\n",
       "      <th>AVG_avg_precisions</th>\n",
       "      <th>flag_binary_support</th>\n",
       "      <th>flag_frequency_support</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>146</th>\n",
       "      <td>kmeans__k=2</td>\n",
       "      <td>transitions</td>\n",
       "      <td>False</td>\n",
       "      <td>cb</td>\n",
       "      <td>125</td>\n",
       "      <td>0.683422</td>\n",
       "      <td>0.569979</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>152</th>\n",
       "      <td>kmeans__k=2</td>\n",
       "      <td>transitions</td>\n",
       "      <td>False</td>\n",
       "      <td>cd</td>\n",
       "      <td>125</td>\n",
       "      <td>0.798775</td>\n",
       "      <td>0.285937</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>158</th>\n",
       "      <td>kmeans__k=2</td>\n",
       "      <td>transitions</td>\n",
       "      <td>False</td>\n",
       "      <td>cf</td>\n",
       "      <td>125</td>\n",
       "      <td>0.979377</td>\n",
       "      <td>0.870062</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>164</th>\n",
       "      <td>kmeans__k=2</td>\n",
       "      <td>transitions</td>\n",
       "      <td>False</td>\n",
       "      <td>cm</td>\n",
       "      <td>125</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>170</th>\n",
       "      <td>kmeans__k=2</td>\n",
       "      <td>transitions</td>\n",
       "      <td>False</td>\n",
       "      <td>cp</td>\n",
       "      <td>125</td>\n",
       "      <td>0.973333</td>\n",
       "      <td>0.835802</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>176</th>\n",
       "      <td>kmeans__k=2</td>\n",
       "      <td>transitions</td>\n",
       "      <td>False</td>\n",
       "      <td>fr</td>\n",
       "      <td>125</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>182</th>\n",
       "      <td>kmeans__k=2</td>\n",
       "      <td>transitions</td>\n",
       "      <td>False</td>\n",
       "      <td>lp</td>\n",
       "      <td>125</td>\n",
       "      <td>0.783273</td>\n",
       "      <td>0.305099</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>188</th>\n",
       "      <td>kmeans__k=2</td>\n",
       "      <td>transitions</td>\n",
       "      <td>False</td>\n",
       "      <td>pl</td>\n",
       "      <td>125</td>\n",
       "      <td>0.914074</td>\n",
       "      <td>0.666843</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>194</th>\n",
       "      <td>kmeans__k=2</td>\n",
       "      <td>transitions</td>\n",
       "      <td>False</td>\n",
       "      <td>pm</td>\n",
       "      <td>125</td>\n",
       "      <td>0.950426</td>\n",
       "      <td>0.757172</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>200</th>\n",
       "      <td>kmeans__k=2</td>\n",
       "      <td>transitions</td>\n",
       "      <td>False</td>\n",
       "      <td>re</td>\n",
       "      <td>125</td>\n",
       "      <td>0.989836</td>\n",
       "      <td>0.762757</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>206</th>\n",
       "      <td>kmeans__k=2</td>\n",
       "      <td>transitions</td>\n",
       "      <td>False</td>\n",
       "      <td>rp</td>\n",
       "      <td>125</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>212</th>\n",
       "      <td>kmeans__k=2</td>\n",
       "      <td>transitions</td>\n",
       "      <td>False</td>\n",
       "      <td>sw</td>\n",
       "      <td>125</td>\n",
       "      <td>0.927273</td>\n",
       "      <td>0.828840</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "           model representation  sliding_window tipo_mudanca  window_size  \\\n",
       "146  kmeans__k=2    transitions           False           cb          125   \n",
       "152  kmeans__k=2    transitions           False           cd          125   \n",
       "158  kmeans__k=2    transitions           False           cf          125   \n",
       "164  kmeans__k=2    transitions           False           cm          125   \n",
       "170  kmeans__k=2    transitions           False           cp          125   \n",
       "176  kmeans__k=2    transitions           False           fr          125   \n",
       "182  kmeans__k=2    transitions           False           lp          125   \n",
       "188  kmeans__k=2    transitions           False           pl          125   \n",
       "194  kmeans__k=2    transitions           False           pm          125   \n",
       "200  kmeans__k=2    transitions           False           re          125   \n",
       "206  kmeans__k=2    transitions           False           rp          125   \n",
       "212  kmeans__k=2    transitions           False           sw          125   \n",
       "\n",
       "     AVG_roc_scores  AVG_avg_precisions  flag_binary_support  \\\n",
       "146        0.683422            0.569979                 True   \n",
       "152        0.798775            0.285937                False   \n",
       "158        0.979377            0.870062                 True   \n",
       "164        1.000000            1.000000                 True   \n",
       "170        0.973333            0.835802                False   \n",
       "176        1.000000            1.000000                False   \n",
       "182        0.783273            0.305099                False   \n",
       "188        0.914074            0.666843                False   \n",
       "194        0.950426            0.757172                 True   \n",
       "200        0.989836            0.762757                 True   \n",
       "206        1.000000            1.000000                 True   \n",
       "212        0.927273            0.828840                False   \n",
       "\n",
       "     flag_frequency_support  \n",
       "146                    True  \n",
       "152                   False  \n",
       "158                    True  \n",
       "164                    True  \n",
       "170                    True  \n",
       "176                   False  \n",
       "182                    True  \n",
       "188                   False  \n",
       "194                    True  \n",
       "200                    True  \n",
       "206                    True  \n",
       "212                   False  "
      ]
     },
     "execution_count": 41,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "resp_df_grouped[\n",
    "    (resp_df_grouped.model == \"kmeans__k=2\")\n",
    "    & (resp_df_grouped.representation == \"transitions\")\n",
    "    & (resp_df_grouped.window_size == 125)\n",
    "]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 533,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>model</th>\n",
       "      <th>representation</th>\n",
       "      <th>sliding_window</th>\n",
       "      <th>window_size</th>\n",
       "      <th>AVG_roc_scores</th>\n",
       "      <th>AVG_avg_precisions</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>kmeans__k=2</td>\n",
       "      <td>binary</td>\n",
       "      <td>False</td>\n",
       "      <td>75</td>\n",
       "      <td>0.998457</td>\n",
       "      <td>0.972222</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>kmeans__k=2</td>\n",
       "      <td>binary</td>\n",
       "      <td>False</td>\n",
       "      <td>100</td>\n",
       "      <td>0.998457</td>\n",
       "      <td>0.972222</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>kmeans__k=2</td>\n",
       "      <td>binary</td>\n",
       "      <td>False</td>\n",
       "      <td>125</td>\n",
       "      <td>0.998457</td>\n",
       "      <td>0.972222</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>kmeans__k=2</td>\n",
       "      <td>binary</td>\n",
       "      <td>False</td>\n",
       "      <td>150</td>\n",
       "      <td>0.998457</td>\n",
       "      <td>0.972222</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>kmeans__k=2</td>\n",
       "      <td>binary</td>\n",
       "      <td>False</td>\n",
       "      <td>200</td>\n",
       "      <td>0.998457</td>\n",
       "      <td>0.972222</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "         model representation  sliding_window  window_size  AVG_roc_scores  \\\n",
       "0  kmeans__k=2         binary           False           75        0.998457   \n",
       "1  kmeans__k=2         binary           False          100        0.998457   \n",
       "2  kmeans__k=2         binary           False          125        0.998457   \n",
       "3  kmeans__k=2         binary           False          150        0.998457   \n",
       "4  kmeans__k=2         binary           False          200        0.998457   \n",
       "\n",
       "   AVG_avg_precisions  \n",
       "0            0.972222  \n",
       "1            0.972222  \n",
       "2            0.972222  \n",
       "3            0.972222  \n",
       "4            0.972222  "
      ]
     },
     "execution_count": 533,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "resp_df_grouped[resp_df_grouped.flag_binary_support].groupby(['model', 'representation', 'sliding_window', 'window_size']).agg({\n",
    "    'AVG_roc_scores': 'mean',\n",
    "    'AVG_avg_precisions': 'mean',\n",
    "}).reset_index().sort_values([\"AVG_roc_scores\", \"AVG_avg_precisions\"], ascending=False).head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 534,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>model</th>\n",
       "      <th>representation</th>\n",
       "      <th>sliding_window</th>\n",
       "      <th>window_size</th>\n",
       "      <th>AVG_roc_scores</th>\n",
       "      <th>AVG_avg_precisions</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>kmeans__k=2</td>\n",
       "      <td>frequency</td>\n",
       "      <td>False</td>\n",
       "      <td>100</td>\n",
       "      <td>0.996219</td>\n",
       "      <td>0.971726</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>kmeans__k=2</td>\n",
       "      <td>frequency</td>\n",
       "      <td>False</td>\n",
       "      <td>200</td>\n",
       "      <td>0.995602</td>\n",
       "      <td>0.966766</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>kmeans__k=2</td>\n",
       "      <td>frequency</td>\n",
       "      <td>False</td>\n",
       "      <td>125</td>\n",
       "      <td>0.995448</td>\n",
       "      <td>0.968254</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>kmeans__k=2</td>\n",
       "      <td>frequency</td>\n",
       "      <td>False</td>\n",
       "      <td>250</td>\n",
       "      <td>0.994522</td>\n",
       "      <td>0.966270</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>26</th>\n",
       "      <td>kmeans__k=3</td>\n",
       "      <td>frequency</td>\n",
       "      <td>False</td>\n",
       "      <td>125</td>\n",
       "      <td>0.993638</td>\n",
       "      <td>0.960648</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "          model representation  sliding_window  window_size  AVG_roc_scores  \\\n",
       "7   kmeans__k=2      frequency           False          100        0.996219   \n",
       "10  kmeans__k=2      frequency           False          200        0.995602   \n",
       "8   kmeans__k=2      frequency           False          125        0.995448   \n",
       "11  kmeans__k=2      frequency           False          250        0.994522   \n",
       "26  kmeans__k=3      frequency           False          125        0.993638   \n",
       "\n",
       "    AVG_avg_precisions  \n",
       "7             0.971726  \n",
       "10            0.966766  \n",
       "8             0.968254  \n",
       "11            0.966270  \n",
       "26            0.960648  "
      ]
     },
     "execution_count": 534,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "resp_df_grouped[resp_df_grouped.flag_frequency_support].groupby(['model', 'representation', 'sliding_window', 'window_size']).agg({\n",
    "    'AVG_roc_scores': 'mean',\n",
    "    'AVG_avg_precisions': 'mean',\n",
    "}).reset_index().sort_values([\"AVG_roc_scores\", \"AVG_avg_precisions\"], ascending=False).head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Plot"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {},
   "outputs": [],
   "source": [
    "log = pm.all_prep(r\"C:\\Users\\rafae\\Google Drive\\USP\\Mestrado\\Experimentos\\process_mining_datasets\\fr\\fr7.5k.MXML\", aliases)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [],
   "source": [
    "combinations = list(\n",
    "    ParameterGrid({\n",
    "        'tipo_mudanca': ['cp'],\n",
    "        'log_size': [5000],\n",
    "        'representation': ['frequency'],\n",
    "        'window_size': [100],\n",
    "        'model': ['kmeans__k=2'],\n",
    "        'sliding_window': [False]\n",
    "    })\n",
    ")\n",
    " \n",
    "c = combinations[0]\n",
    "\n",
    "df = pd.read_csv(base_dir + c[\"tipo_mudanca\"] + \"/\" + opts_to_string(combinations[0]) + \".csv\", index_col=0)\n",
    "df.centroids = df.centroids.apply(string_to_numpy)\n",
    "df.index += df.index[1]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [],
   "source": [
    "v = []\n",
    "previous = df.centroids.iloc[0]\n",
    "for i in range(1, len(df.centroids)):\n",
    "    v.append(\n",
    "        localize_drift(previous, df.centroids.iloc[i], cached_resp[\"_\".join([c[\"tipo_mudanca\"], str(c[\"log_size\"]), c[\"representation\"]])][0]).to_dict()[0]\n",
    "    )\n",
    "    previous = df.centroids.iloc[i]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [],
   "source": [
    "df_plot = pd.DataFrame(v)\n",
    "df_plot.index = df.index[1:]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [],
   "source": [
    "df_plot.loc[c['window_size']] = 0\n",
    "df_plot.loc[0] = 0\n",
    "df_plot = df_plot.sort_index()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [],
   "source": [
    "def reinverse_transitions(df):\n",
    "    new_cols = []\n",
    "    for col in df.columns:\n",
    "        a, b = col.split(\" \")\n",
    "        new_cols.append(\n",
    "            inv_aliases[a] + \" -> \" + inv_aliases[b]\n",
    "        )\n",
    "    df.columns = new_cols"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [],
   "source": [
    "if c[\"representation\"] == 'transitions':\n",
    "    reinverse_transitions(df_plot)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>A</th>\n",
       "      <th>B</th>\n",
       "      <th>C</th>\n",
       "      <th>D</th>\n",
       "      <th>E</th>\n",
       "      <th>END_A</th>\n",
       "      <th>END_C</th>\n",
       "      <th>END_R</th>\n",
       "      <th>F</th>\n",
       "      <th>G</th>\n",
       "      <th>H</th>\n",
       "      <th>I</th>\n",
       "      <th>J</th>\n",
       "      <th>K</th>\n",
       "      <th>L</th>\n",
       "      <th>M</th>\n",
       "      <th>N</th>\n",
       "      <th>O</th>\n",
       "      <th>START</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>i</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>100</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>200</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000058</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000058</td>\n",
       "      <td>0.000058</td>\n",
       "      <td>0.000058</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.003194</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.003194</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.003194</td>\n",
       "      <td>0.012151</td>\n",
       "      <td>0.012151</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>300</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.005253</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.005253</td>\n",
       "      <td>0.005253</td>\n",
       "      <td>0.005253</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.014333</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.014333</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.014333</td>\n",
       "      <td>0.010392</td>\n",
       "      <td>0.010392</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>400</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.002689</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.002689</td>\n",
       "      <td>0.002689</td>\n",
       "      <td>0.002689</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.010000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.010000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.010000</td>\n",
       "      <td>0.004565</td>\n",
       "      <td>0.004565</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>500</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.001467</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.001467</td>\n",
       "      <td>0.001467</td>\n",
       "      <td>0.001467</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.014672</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.014672</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.014672</td>\n",
       "      <td>0.024691</td>\n",
       "      <td>0.024691</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>600</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.002620</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.5</td>\n",
       "      <td>0.002620</td>\n",
       "      <td>0.002620</td>\n",
       "      <td>0.002620</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.004850</td>\n",
       "      <td>0.5</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.004850</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.004850</td>\n",
       "      <td>0.001596</td>\n",
       "      <td>0.001596</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>700</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000786</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000786</td>\n",
       "      <td>0.000786</td>\n",
       "      <td>0.000786</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.004297</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.004297</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.004297</td>\n",
       "      <td>0.006069</td>\n",
       "      <td>0.006069</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>800</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.002061</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.002061</td>\n",
       "      <td>0.002061</td>\n",
       "      <td>0.002061</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.005445</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.005445</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.005445</td>\n",
       "      <td>0.001415</td>\n",
       "      <td>0.001415</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>900</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.002365</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.002365</td>\n",
       "      <td>0.002365</td>\n",
       "      <td>0.002365</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.001666</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.001666</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.001666</td>\n",
       "      <td>0.002101</td>\n",
       "      <td>0.002101</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1000</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.005366</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.005366</td>\n",
       "      <td>0.005366</td>\n",
       "      <td>0.005366</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000140</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000140</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000140</td>\n",
       "      <td>0.010951</td>\n",
       "      <td>0.010951</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1100</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000212</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.5</td>\n",
       "      <td>0.000212</td>\n",
       "      <td>0.000212</td>\n",
       "      <td>0.000212</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.005863</td>\n",
       "      <td>0.5</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.005863</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.005863</td>\n",
       "      <td>0.016842</td>\n",
       "      <td>0.016842</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1200</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000011</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000011</td>\n",
       "      <td>0.000011</td>\n",
       "      <td>0.000011</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.007110</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.007110</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.007110</td>\n",
       "      <td>0.000101</td>\n",
       "      <td>0.000101</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1300</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.002535</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.002535</td>\n",
       "      <td>0.002535</td>\n",
       "      <td>0.002535</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000813</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000813</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000813</td>\n",
       "      <td>0.000360</td>\n",
       "      <td>0.000360</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1400</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.001419</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.001419</td>\n",
       "      <td>0.001419</td>\n",
       "      <td>0.001419</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000287</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000287</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000287</td>\n",
       "      <td>0.011774</td>\n",
       "      <td>0.011774</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1500</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.002928</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.002928</td>\n",
       "      <td>0.002928</td>\n",
       "      <td>0.002928</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.012465</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.012465</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.012465</td>\n",
       "      <td>0.001022</td>\n",
       "      <td>0.001022</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1600</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.004731</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.5</td>\n",
       "      <td>0.004731</td>\n",
       "      <td>0.004731</td>\n",
       "      <td>0.004731</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000143</td>\n",
       "      <td>0.5</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000143</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000143</td>\n",
       "      <td>0.006150</td>\n",
       "      <td>0.006150</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1700</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000084</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000084</td>\n",
       "      <td>0.000084</td>\n",
       "      <td>0.000084</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.001851</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.001851</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.001851</td>\n",
       "      <td>0.007239</td>\n",
       "      <td>0.007239</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1800</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.001361</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.001361</td>\n",
       "      <td>0.001361</td>\n",
       "      <td>0.001361</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.002571</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.002571</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.002571</td>\n",
       "      <td>0.007519</td>\n",
       "      <td>0.007519</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1900</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.004601</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.004601</td>\n",
       "      <td>0.004601</td>\n",
       "      <td>0.004601</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000788</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000788</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000788</td>\n",
       "      <td>0.001879</td>\n",
       "      <td>0.001879</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2000</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.002670</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.002670</td>\n",
       "      <td>0.002670</td>\n",
       "      <td>0.002670</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000778</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000778</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000778</td>\n",
       "      <td>0.000509</td>\n",
       "      <td>0.000509</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2100</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.001022</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.5</td>\n",
       "      <td>0.001022</td>\n",
       "      <td>0.001022</td>\n",
       "      <td>0.001022</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.003793</td>\n",
       "      <td>0.5</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.003793</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.003793</td>\n",
       "      <td>0.001800</td>\n",
       "      <td>0.001800</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2200</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000381</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000381</td>\n",
       "      <td>0.000381</td>\n",
       "      <td>0.000381</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.002241</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.002241</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.002241</td>\n",
       "      <td>0.000655</td>\n",
       "      <td>0.000655</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2300</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000283</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000283</td>\n",
       "      <td>0.000283</td>\n",
       "      <td>0.000283</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.006862</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.006862</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.006862</td>\n",
       "      <td>0.006687</td>\n",
       "      <td>0.006687</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2400</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.004973</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.004973</td>\n",
       "      <td>0.004973</td>\n",
       "      <td>0.004973</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.003399</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.003399</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.003399</td>\n",
       "      <td>0.003395</td>\n",
       "      <td>0.003395</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2500</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.001602</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.001602</td>\n",
       "      <td>0.001602</td>\n",
       "      <td>0.001602</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.014664</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.014664</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.014664</td>\n",
       "      <td>0.000712</td>\n",
       "      <td>0.000712</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2600</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.002790</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.5</td>\n",
       "      <td>0.002790</td>\n",
       "      <td>0.002790</td>\n",
       "      <td>0.002790</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.001244</td>\n",
       "      <td>0.5</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.001244</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.001244</td>\n",
       "      <td>0.016168</td>\n",
       "      <td>0.016168</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2700</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.006589</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.006589</td>\n",
       "      <td>0.006589</td>\n",
       "      <td>0.006589</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.001061</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.001061</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.001061</td>\n",
       "      <td>0.028042</td>\n",
       "      <td>0.028042</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2800</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.005331</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.005331</td>\n",
       "      <td>0.005331</td>\n",
       "      <td>0.005331</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.001146</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.001146</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.001146</td>\n",
       "      <td>0.003397</td>\n",
       "      <td>0.003397</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2900</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.002055</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.002055</td>\n",
       "      <td>0.002055</td>\n",
       "      <td>0.002055</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.006995</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.006995</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.006995</td>\n",
       "      <td>0.003285</td>\n",
       "      <td>0.003285</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3000</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.003188</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.003188</td>\n",
       "      <td>0.003188</td>\n",
       "      <td>0.003188</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.007340</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.007340</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.007340</td>\n",
       "      <td>0.002480</td>\n",
       "      <td>0.002480</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3100</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.003100</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.5</td>\n",
       "      <td>0.003100</td>\n",
       "      <td>0.003100</td>\n",
       "      <td>0.003100</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.004447</td>\n",
       "      <td>0.5</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.004447</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.004447</td>\n",
       "      <td>0.007119</td>\n",
       "      <td>0.007119</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3200</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000115</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000115</td>\n",
       "      <td>0.000115</td>\n",
       "      <td>0.000115</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000003</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000003</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000003</td>\n",
       "      <td>0.009782</td>\n",
       "      <td>0.009782</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3300</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000010</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000010</td>\n",
       "      <td>0.000010</td>\n",
       "      <td>0.000010</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.005333</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.005333</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.005333</td>\n",
       "      <td>0.001922</td>\n",
       "      <td>0.001922</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3400</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.002399</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.002399</td>\n",
       "      <td>0.002399</td>\n",
       "      <td>0.002399</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.003313</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.003313</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.003313</td>\n",
       "      <td>0.001186</td>\n",
       "      <td>0.001186</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3500</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000649</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000649</td>\n",
       "      <td>0.000649</td>\n",
       "      <td>0.000649</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.001222</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.001222</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.001222</td>\n",
       "      <td>0.000088</td>\n",
       "      <td>0.000088</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3600</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.003005</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.5</td>\n",
       "      <td>0.003005</td>\n",
       "      <td>0.003005</td>\n",
       "      <td>0.003005</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.003058</td>\n",
       "      <td>0.5</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.003058</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.003058</td>\n",
       "      <td>0.003005</td>\n",
       "      <td>0.003005</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3700</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.002912</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.002912</td>\n",
       "      <td>0.002912</td>\n",
       "      <td>0.002912</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000201</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000201</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000201</td>\n",
       "      <td>0.002912</td>\n",
       "      <td>0.002912</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3800</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000737</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000737</td>\n",
       "      <td>0.000737</td>\n",
       "      <td>0.000737</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.001099</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.001099</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.001099</td>\n",
       "      <td>0.000209</td>\n",
       "      <td>0.000209</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3900</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.013228</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.013228</td>\n",
       "      <td>0.013228</td>\n",
       "      <td>0.013228</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.001339</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.001339</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.001339</td>\n",
       "      <td>0.003041</td>\n",
       "      <td>0.003041</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4000</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.014313</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.014313</td>\n",
       "      <td>0.014313</td>\n",
       "      <td>0.014313</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.001349</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.001349</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.001349</td>\n",
       "      <td>0.005536</td>\n",
       "      <td>0.005536</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4100</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.008062</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.5</td>\n",
       "      <td>0.008062</td>\n",
       "      <td>0.008062</td>\n",
       "      <td>0.008062</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.001156</td>\n",
       "      <td>0.5</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.001156</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.001156</td>\n",
       "      <td>0.002828</td>\n",
       "      <td>0.002828</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4200</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.004110</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.004110</td>\n",
       "      <td>0.004110</td>\n",
       "      <td>0.004110</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.014910</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.014910</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.014910</td>\n",
       "      <td>0.035402</td>\n",
       "      <td>0.035402</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4300</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.001130</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.001130</td>\n",
       "      <td>0.001130</td>\n",
       "      <td>0.001130</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.024414</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.024414</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.024414</td>\n",
       "      <td>0.005690</td>\n",
       "      <td>0.005690</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4400</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.003146</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.003146</td>\n",
       "      <td>0.003146</td>\n",
       "      <td>0.003146</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.003097</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.003097</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.003097</td>\n",
       "      <td>0.000137</td>\n",
       "      <td>0.000137</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4500</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.011404</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.011404</td>\n",
       "      <td>0.011404</td>\n",
       "      <td>0.011404</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.002672</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.002672</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.002672</td>\n",
       "      <td>0.000006</td>\n",
       "      <td>0.000006</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4600</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.001800</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.5</td>\n",
       "      <td>0.001800</td>\n",
       "      <td>0.001800</td>\n",
       "      <td>0.001800</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000099</td>\n",
       "      <td>0.5</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000099</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000099</td>\n",
       "      <td>0.000026</td>\n",
       "      <td>0.000026</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4700</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.004492</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.004492</td>\n",
       "      <td>0.004492</td>\n",
       "      <td>0.004492</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000593</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000593</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000593</td>\n",
       "      <td>0.000200</td>\n",
       "      <td>0.000200</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4800</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.012954</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.012954</td>\n",
       "      <td>0.012954</td>\n",
       "      <td>0.012954</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000892</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000892</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000892</td>\n",
       "      <td>0.016354</td>\n",
       "      <td>0.016354</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4900</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.007832</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.007832</td>\n",
       "      <td>0.007832</td>\n",
       "      <td>0.007832</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.003810</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.003810</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.003810</td>\n",
       "      <td>0.004289</td>\n",
       "      <td>0.004289</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5000</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.005103</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.005103</td>\n",
       "      <td>0.005103</td>\n",
       "      <td>0.005103</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.011217</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.011217</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.011217</td>\n",
       "      <td>0.011774</td>\n",
       "      <td>0.011774</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "        A         B    C    D         E     END_A     END_C  END_R         F  \\\n",
       "i                                                                              \n",
       "0     0.0  0.000000  0.0  0.0  0.000000  0.000000  0.000000    0.0  0.000000   \n",
       "100   0.0  0.000000  0.0  0.0  0.000000  0.000000  0.000000    0.0  0.000000   \n",
       "200   0.0  0.000058  0.0  0.0  0.000058  0.000058  0.000058    0.0  0.003194   \n",
       "300   0.0  0.005253  0.0  0.0  0.005253  0.005253  0.005253    0.0  0.014333   \n",
       "400   0.0  0.002689  0.0  0.0  0.002689  0.002689  0.002689    0.0  0.010000   \n",
       "500   0.0  0.001467  0.0  0.0  0.001467  0.001467  0.001467    0.0  0.014672   \n",
       "600   0.0  0.002620  0.0  0.5  0.002620  0.002620  0.002620    0.0  0.004850   \n",
       "700   0.0  0.000786  0.0  0.0  0.000786  0.000786  0.000786    0.0  0.004297   \n",
       "800   0.0  0.002061  0.0  0.0  0.002061  0.002061  0.002061    0.0  0.005445   \n",
       "900   0.0  0.002365  0.0  0.0  0.002365  0.002365  0.002365    0.0  0.001666   \n",
       "1000  0.0  0.005366  0.0  0.0  0.005366  0.005366  0.005366    0.0  0.000140   \n",
       "1100  0.0  0.000212  0.0  0.5  0.000212  0.000212  0.000212    0.0  0.005863   \n",
       "1200  0.0  0.000011  0.0  0.0  0.000011  0.000011  0.000011    0.0  0.007110   \n",
       "1300  0.0  0.002535  0.0  0.0  0.002535  0.002535  0.002535    0.0  0.000813   \n",
       "1400  0.0  0.001419  0.0  0.0  0.001419  0.001419  0.001419    0.0  0.000287   \n",
       "1500  0.0  0.002928  0.0  0.0  0.002928  0.002928  0.002928    0.0  0.012465   \n",
       "1600  0.0  0.004731  0.0  0.5  0.004731  0.004731  0.004731    0.0  0.000143   \n",
       "1700  0.0  0.000084  0.0  0.0  0.000084  0.000084  0.000084    0.0  0.001851   \n",
       "1800  0.0  0.001361  0.0  0.0  0.001361  0.001361  0.001361    0.0  0.002571   \n",
       "1900  0.0  0.004601  0.0  0.0  0.004601  0.004601  0.004601    0.0  0.000788   \n",
       "2000  0.0  0.002670  0.0  0.0  0.002670  0.002670  0.002670    0.0  0.000778   \n",
       "2100  0.0  0.001022  0.0  0.5  0.001022  0.001022  0.001022    0.0  0.003793   \n",
       "2200  0.0  0.000381  0.0  0.0  0.000381  0.000381  0.000381    0.0  0.002241   \n",
       "2300  0.0  0.000283  0.0  0.0  0.000283  0.000283  0.000283    0.0  0.006862   \n",
       "2400  0.0  0.004973  0.0  0.0  0.004973  0.004973  0.004973    0.0  0.003399   \n",
       "2500  0.0  0.001602  0.0  0.0  0.001602  0.001602  0.001602    0.0  0.014664   \n",
       "2600  0.0  0.002790  0.0  0.5  0.002790  0.002790  0.002790    0.0  0.001244   \n",
       "2700  0.0  0.006589  0.0  0.0  0.006589  0.006589  0.006589    0.0  0.001061   \n",
       "2800  0.0  0.005331  0.0  0.0  0.005331  0.005331  0.005331    0.0  0.001146   \n",
       "2900  0.0  0.002055  0.0  0.0  0.002055  0.002055  0.002055    0.0  0.006995   \n",
       "3000  0.0  0.003188  0.0  0.0  0.003188  0.003188  0.003188    0.0  0.007340   \n",
       "3100  0.0  0.003100  0.0  0.5  0.003100  0.003100  0.003100    0.0  0.004447   \n",
       "3200  0.0  0.000115  0.0  0.0  0.000115  0.000115  0.000115    0.0  0.000003   \n",
       "3300  0.0  0.000010  0.0  0.0  0.000010  0.000010  0.000010    0.0  0.005333   \n",
       "3400  0.0  0.002399  0.0  0.0  0.002399  0.002399  0.002399    0.0  0.003313   \n",
       "3500  0.0  0.000649  0.0  0.0  0.000649  0.000649  0.000649    0.0  0.001222   \n",
       "3600  0.0  0.003005  0.0  0.5  0.003005  0.003005  0.003005    0.0  0.003058   \n",
       "3700  0.0  0.002912  0.0  0.0  0.002912  0.002912  0.002912    0.0  0.000201   \n",
       "3800  0.0  0.000737  0.0  0.0  0.000737  0.000737  0.000737    0.0  0.001099   \n",
       "3900  0.0  0.013228  0.0  0.0  0.013228  0.013228  0.013228    0.0  0.001339   \n",
       "4000  0.0  0.014313  0.0  0.0  0.014313  0.014313  0.014313    0.0  0.001349   \n",
       "4100  0.0  0.008062  0.0  0.5  0.008062  0.008062  0.008062    0.0  0.001156   \n",
       "4200  0.0  0.004110  0.0  0.0  0.004110  0.004110  0.004110    0.0  0.014910   \n",
       "4300  0.0  0.001130  0.0  0.0  0.001130  0.001130  0.001130    0.0  0.024414   \n",
       "4400  0.0  0.003146  0.0  0.0  0.003146  0.003146  0.003146    0.0  0.003097   \n",
       "4500  0.0  0.011404  0.0  0.0  0.011404  0.011404  0.011404    0.0  0.002672   \n",
       "4600  0.0  0.001800  0.0  0.5  0.001800  0.001800  0.001800    0.0  0.000099   \n",
       "4700  0.0  0.004492  0.0  0.0  0.004492  0.004492  0.004492    0.0  0.000593   \n",
       "4800  0.0  0.012954  0.0  0.0  0.012954  0.012954  0.012954    0.0  0.000892   \n",
       "4900  0.0  0.007832  0.0  0.0  0.007832  0.007832  0.007832    0.0  0.003810   \n",
       "5000  0.0  0.005103  0.0  0.0  0.005103  0.005103  0.005103    0.0  0.011217   \n",
       "\n",
       "        G    H    I         J    K         L         M         N    O  START  \n",
       "i                                                                             \n",
       "0     0.0  0.0  0.0  0.000000  0.0  0.000000  0.000000  0.000000  0.0    0.0  \n",
       "100   0.0  0.0  0.0  0.000000  0.0  0.000000  0.000000  0.000000  0.0    0.0  \n",
       "200   0.0  0.0  0.0  0.003194  0.0  0.003194  0.012151  0.012151  0.0    0.0  \n",
       "300   0.0  0.0  0.0  0.014333  0.0  0.014333  0.010392  0.010392  0.0    0.0  \n",
       "400   0.0  0.0  0.0  0.010000  0.0  0.010000  0.004565  0.004565  0.0    0.0  \n",
       "500   0.0  0.0  0.0  0.014672  0.0  0.014672  0.024691  0.024691  0.0    0.0  \n",
       "600   0.5  0.0  0.0  0.004850  0.0  0.004850  0.001596  0.001596  0.0    0.0  \n",
       "700   0.0  0.0  0.0  0.004297  0.0  0.004297  0.006069  0.006069  0.0    0.0  \n",
       "800   0.0  0.0  0.0  0.005445  0.0  0.005445  0.001415  0.001415  0.0    0.0  \n",
       "900   0.0  0.0  0.0  0.001666  0.0  0.001666  0.002101  0.002101  0.0    0.0  \n",
       "1000  0.0  0.0  0.0  0.000140  0.0  0.000140  0.010951  0.010951  0.0    0.0  \n",
       "1100  0.5  0.0  0.0  0.005863  0.0  0.005863  0.016842  0.016842  0.0    0.0  \n",
       "1200  0.0  0.0  0.0  0.007110  0.0  0.007110  0.000101  0.000101  0.0    0.0  \n",
       "1300  0.0  0.0  0.0  0.000813  0.0  0.000813  0.000360  0.000360  0.0    0.0  \n",
       "1400  0.0  0.0  0.0  0.000287  0.0  0.000287  0.011774  0.011774  0.0    0.0  \n",
       "1500  0.0  0.0  0.0  0.012465  0.0  0.012465  0.001022  0.001022  0.0    0.0  \n",
       "1600  0.5  0.0  0.0  0.000143  0.0  0.000143  0.006150  0.006150  0.0    0.0  \n",
       "1700  0.0  0.0  0.0  0.001851  0.0  0.001851  0.007239  0.007239  0.0    0.0  \n",
       "1800  0.0  0.0  0.0  0.002571  0.0  0.002571  0.007519  0.007519  0.0    0.0  \n",
       "1900  0.0  0.0  0.0  0.000788  0.0  0.000788  0.001879  0.001879  0.0    0.0  \n",
       "2000  0.0  0.0  0.0  0.000778  0.0  0.000778  0.000509  0.000509  0.0    0.0  \n",
       "2100  0.5  0.0  0.0  0.003793  0.0  0.003793  0.001800  0.001800  0.0    0.0  \n",
       "2200  0.0  0.0  0.0  0.002241  0.0  0.002241  0.000655  0.000655  0.0    0.0  \n",
       "2300  0.0  0.0  0.0  0.006862  0.0  0.006862  0.006687  0.006687  0.0    0.0  \n",
       "2400  0.0  0.0  0.0  0.003399  0.0  0.003399  0.003395  0.003395  0.0    0.0  \n",
       "2500  0.0  0.0  0.0  0.014664  0.0  0.014664  0.000712  0.000712  0.0    0.0  \n",
       "2600  0.5  0.0  0.0  0.001244  0.0  0.001244  0.016168  0.016168  0.0    0.0  \n",
       "2700  0.0  0.0  0.0  0.001061  0.0  0.001061  0.028042  0.028042  0.0    0.0  \n",
       "2800  0.0  0.0  0.0  0.001146  0.0  0.001146  0.003397  0.003397  0.0    0.0  \n",
       "2900  0.0  0.0  0.0  0.006995  0.0  0.006995  0.003285  0.003285  0.0    0.0  \n",
       "3000  0.0  0.0  0.0  0.007340  0.0  0.007340  0.002480  0.002480  0.0    0.0  \n",
       "3100  0.5  0.0  0.0  0.004447  0.0  0.004447  0.007119  0.007119  0.0    0.0  \n",
       "3200  0.0  0.0  0.0  0.000003  0.0  0.000003  0.009782  0.009782  0.0    0.0  \n",
       "3300  0.0  0.0  0.0  0.005333  0.0  0.005333  0.001922  0.001922  0.0    0.0  \n",
       "3400  0.0  0.0  0.0  0.003313  0.0  0.003313  0.001186  0.001186  0.0    0.0  \n",
       "3500  0.0  0.0  0.0  0.001222  0.0  0.001222  0.000088  0.000088  0.0    0.0  \n",
       "3600  0.5  0.0  0.0  0.003058  0.0  0.003058  0.003005  0.003005  0.0    0.0  \n",
       "3700  0.0  0.0  0.0  0.000201  0.0  0.000201  0.002912  0.002912  0.0    0.0  \n",
       "3800  0.0  0.0  0.0  0.001099  0.0  0.001099  0.000209  0.000209  0.0    0.0  \n",
       "3900  0.0  0.0  0.0  0.001339  0.0  0.001339  0.003041  0.003041  0.0    0.0  \n",
       "4000  0.0  0.0  0.0  0.001349  0.0  0.001349  0.005536  0.005536  0.0    0.0  \n",
       "4100  0.5  0.0  0.0  0.001156  0.0  0.001156  0.002828  0.002828  0.0    0.0  \n",
       "4200  0.0  0.0  0.0  0.014910  0.0  0.014910  0.035402  0.035402  0.0    0.0  \n",
       "4300  0.0  0.0  0.0  0.024414  0.0  0.024414  0.005690  0.005690  0.0    0.0  \n",
       "4400  0.0  0.0  0.0  0.003097  0.0  0.003097  0.000137  0.000137  0.0    0.0  \n",
       "4500  0.0  0.0  0.0  0.002672  0.0  0.002672  0.000006  0.000006  0.0    0.0  \n",
       "4600  0.5  0.0  0.0  0.000099  0.0  0.000099  0.000026  0.000026  0.0    0.0  \n",
       "4700  0.0  0.0  0.0  0.000593  0.0  0.000593  0.000200  0.000200  0.0    0.0  \n",
       "4800  0.0  0.0  0.0  0.000892  0.0  0.000892  0.016354  0.016354  0.0    0.0  \n",
       "4900  0.0  0.0  0.0  0.003810  0.0  0.003810  0.004289  0.004289  0.0    0.0  \n",
       "5000  0.0  0.0  0.0  0.011217  0.0  0.011217  0.011774  0.011774  0.0    0.0  "
      ]
     },
     "execution_count": 35,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_plot"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[500, 1000, 1500, 2000, 2500, 3000, 3500, 4000, 4500]"
      ]
     },
     "execution_count": 36,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "list(range(int(c[\"log_size\"]/10), c[\"log_size\"], int(c[\"log_size\"]/10)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 189,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Text(0.5, 0, 'MSE')"
      ]
     },
     "execution_count": 189,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAA3YAAAICCAYAAACDaV/uAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4xLjAsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+17YcXAAAgAElEQVR4nOydd9gcVfXHP9+EEAJJCB0MJUgT6RIUpYgU6VJUQJQqKCqiFKVICYqCFEFpgvwAaaJU6QhSFAQkoSMtdERqIBBCEsr5/XHuZOedzO6+u+++776bnM/z7LOzc8ucuffO7D23nCMzIwiCIAiCIAiCIOhcBrRbgCAIgiAIgiAIgqBnhGIXBEEQBEEQBEHQ4YRiFwRBEARBEARB0OGEYhcEQRAEQRAEQdDhhGIXBEEQBEEQBEHQ4YRiFwRBEARBEARB0OHM1m4BgiAIgiAI+gvzzz+/jRo1qt1iBEEwkzFu3Lg3zGyB3rxGKHZBEARBEASJUaNGMXbs2HaLEQTBTIak53v7GrEUMwiCIAiCIAiCoMMJxS4IgiAIgiAIgqDDCcUuCIIgCIIgCIKgwwnFLgiCIAiCIAiCoMMJxS4IgiAIgiAIgqDDCcUuCIIgCIIgCIKgwwnFLgiCIAiCIAiCoMMJxS4IgiAIgiAIgqDDCcUuCIIgCIIgCIKgw5GZtVuGIAiCIAiCfsEwjbDRrN1uMfoFt9o17RYhCGYaJI0zs9G9eY2YsQuCIAiCoOOQNEjStpKul/R0u+UJgiBoN6HYBUFQE0kHSbIan2mS3pA0VtIpklZot8xBEMzcSNoRuB24DNgEGNheiYIgCNpPKHZBENTj18DcwHbAlNz5l4Drgb8Bk4HVgR8AD0k6QZL6WtAgqIakjdotQ9A6zOwiM/sCcFW7ZQmCIOgvhGIXBEFNzHnHzC4B/pEL2snMNjOzLYBRwLeBqfh7ZT/gpD4XNghKkDQAOK3dcgS9whPtFiAIgqC/EIpdEASN8FrZSTP72MzOBg7Pnd5H0mp9I1YQ1OSbwNLtFiLoFabUjxIEQTBrEIpdEASN8FGd8NOB93O/t+pFWYKgLpJGEbPHMzMft1uAIAiC/kIodkEQtAwzexd4MnfqE+2SJQgkLQXcBMzbblmCIAiCoLcJxS4IglYzNXf8FoCk4ZL2lvSopDHp3HckvSjpZUlb5jOQNLukH0m6W9JESZMlPZgsdM5RiDtA0uaSrpX093RupKSzJL0iaZKk2yRtUE1gSetIujzFnybpdUl3JBlnMAIjaUFJh0h6XtKucg6X9Jqk8ZI+V4g/TNJhkh5I8kySdI+kvdL+r5Yi6SuSbkjWSj+Q9KqkmyR9vU66hSQdL+lxSe+nsv+7pM1rpBksad90P++kdPdL2kdSqaXCVGe7SLpV0gRJU9I1fyVpREn8dSVdLOnJ9HtEMtDzUmob/5a0fSHNlsB95JZgFqy5rpo7P3/KL7vvSZKekHSSpIVL5JlD0s7pns9N5+bJyfRuao+japTbAEm7S7pd0lupDB6TdISkITXSbSXputRGp0p6Rm6NtqWDKJJWk3SmpEm5ez5M0tOpjB6W9P2y5yPFHyXpjBR/ampLD0s6StKwGtdtqlxK8vmayq34jmm4MIIgCDqEUOyCIGgZcqXr07lT/5J0BvAycHIWJmkf4AxgUWAR4KhcHosA9wArAbsBiwE7APMARwN3SBqe4n4PnyG8BtgMGCifpRkL7AosCMwFfBG4SdKuJTJ/BzebvhiwAbAQsA+wSpLx17m4C0i6CHgR+CWweAo6ATgSWABYCjgkl2Z54AFgTmDbdM/fB5bBl65eLWm2mgXbAJKOAv4KGLAGXr5HAl8C/iLp+1XSfRk3RDEU2BQvux8A6wHXSPpZSZolgfuBbdI9LZKuszDw27J7S536G/Gy2A83vLMJMA04GLhP0qIp7tclPYjXz/bA7JLmA/4F7JuuNyTd58WS8ns8rwXmw436ZAzKPmb2QLrGKLx+foDvEV0IWAt4E/gR3obnysl/KPAw8Efgs+ncp4FxwC54PQ/F2+ONZXUraf50T/sBR+BtYg18qfMYvI0PK6QZJOl84AC87S0FrA28kGR/UNIqxWs1iqT1Jd2BK8V7AnOl5/pvwM+BJYA5gBWBU4H/K8ljdbxMvwbshT8Xm+Fl/zPgbyoZ0GimXGpweboWwHP4u2RRcu+aIAiCmY1Q7IIgaCX7451agEdxhWsM8GVc0QBX7r6Ad6BPxmf4/greeU3H48xsDzN7LFnkvArYIqVfHfhNOr4Q+AwwPv0eBJwD/BTvYM9NpSMn4HRJ+Rmc+YHfpbAxZvaomb1lZn/CFROAvVWZeZqIK3075+55w3TdkcCf0v1cnfIfAVwHnGNmB5vZM2b2tpmdhyue4B3eg2qUabeRtBKVzuwBZvasmb1hZqcBF6fzPy5JtwZuNv4GM9srpXvXzC4A7kjRfiFpmVyaEcDNwGBgUzMbZ2bvmdndwJkp2qa5+8w4D28L25jZ/al+b8PbyPvAksC5Ke6NwOeAf+bSn5XyH4Yr7T8EPkxhYyStA9MN+nxIbg+WmX2YfXL5HY/X3dVm9pckz4NU6nhJuu4VPQlYGZiQfo8CTgH2MLP5zWxe4LgUtiywdf7mJc2Ot48VgI3M7LZUbg/jAwTgbfrAQrkdB6wKfNnM7kxy3ou3n5eA+YE/t2CQ4F5gfeCi3LkT8DqYD1fqvgFMSmG7SfpmIY8z8GfvbDO7Kcl6J7B3Cl8zfabTg3Kpxtx42V8HrGpm55rZfwt1HwRBMFMRil0QBM2ycKbwyJdajsFH9ME7mtua2Udm9j8z+xfwegpbC9jTzF4zs32AIWaWzbTsgo/Q/5YCZvZQLo+dJM2VOozv4J1R8A73T83sfDOblpSTw6iYup8Dn/HI+CSumEBaNpojy3MI3mkm5fkGXX1nrQD82MxeNrMd0/2clcIOwGcCTyneD64UZexVEt4My+eOq93PYvmTaSndH/ByOJwZyRQ74cp4xs/x8vu1mb1XJQ347F12rQ3xzvbJZtbF6IWZvQI8kn5uIGmZVL9TcvmNBP5oZielTv8UMzuFygypqCi23SWbYe5SXmY2Hng7/Vwsd36Smb0PPJ1OzYMrqbfkkh9DZSBjjcL19saVmt+b2f8KYXfm0uXLbVlcgT0zXTsv52TgrvRzOXzGtGnSMzMNnznL+I+Z/czMJiTF+GJg91z4oYVsSsuUShuEQjukiXKphqTFcEX0QeArZjaxG2m+I2mspLEfMK1e9CAIgn5Jy5b/BEEwy7E/8Ku0LHIEMBDv7F4GHJcUoDzZ3rtzkpEVwP3k5eLskr5vqrJ1Z770PTveib0v/c56YuPSjFGRMcAeKV1+v9j9wF+S7OMKad7NHQ8uhOX3EZ5qZtOthZbcj4DHq9xPxkhJ85rZhFqRusFNwA3AC2b2ciEsu5/ivayFLzt9wcyeZEaOBJ4C3jOzO8D31VFZ4nhTMYGZ3STfl7cYcEEuKKvfsyWVWVjNGzlZKV0XKvX7XzO7siTdSbgSvSCwvqQhRQWoBsfjbeOckrB38bZdLDOotIH7i4qDmU2Q9Dau9BX3DGZLYcvK7UlJ6+IzcxfngnbCB2KPLFsSi89OZaxUlncTTNduzOzUElkvkfRAkvVTkpYys0zZ/SWwMXBJIVmtZ6qZcpkBuYuVa3Al+MhacQvXOJM00zxMI6xO9CAIgn5JKHZBEDTLT8zstrR8cjgwycym1oifzdCULoVKs3/Z7MYawAd1rv9mdwU1s9cl/QvfL7aopKFp5uUDfO9WXo5V8L1F+eV3XVY3mNnHOUWt2v0siu/peRXvkNbj7fpRamNmb+HLH/NyrI0rLptlpwrJvpi+X6qS5zQqSyMzRuNLXWulu67k9OfT91bAM2XpcnS7PMzsA0nX4cs+B+H7zx6pmaiS9mzg7Oy3pHnwZZjfwvfwQfnqlnquP97BFbvpCoykkUk2qF5ud9B1xhMq5fY9us6klfFunfBWciWVtr08aRbTzH6JK3fA9H2zuwPb5dIOyIU3Wy5dkLQZ8Gfg9EaUuiAIgpmFUOyCIOgRSTnqtpJVg3wn+IO0NK+VPIorduAzHJPygWmG6SB8puIUvNPak5mPbMnY3MCrhZm8XiUZpvgGPov1HL4U9XZyCkyOkel7UAOXGJk7HkR9JTwjK5OPe6l+M+auGqsKyYjKQfj+snPwPZ334MZCWkWx3LpLVm7qhXLrCTXLPO35PAQ3tPJ7YF3KFfZmyyXPbPgS7qHArpJONbPnm8wrCIKgI4k9dkEQ9BfyHboeW/crYWLZsaRFJF2PW/c7ysw2MLMrqDIT1wDZ/cyBG9HoE9J+rLvx/XLfMbNtzOwmKnuUimT/A59s4DL5/45G0mVl0mf1W49kXv8nwOO40rGSmR1tZq+2WkD6Z7n1hGrP1OySjset0/4bWMXMTq2x163ZcsnzIW5pdRJuhfOqvDXTIAiCWYFQ7IIg6C9MoLJcc5taESV9SlKjTqez2cBXzSzzzTU/bmRhI2BjM7uxwTxrkd9jWO9+Rqd9az0iWa38F8mIRrKaWI9MgZkv7U+qlf+6hTTgZdedNFApk3rlsUhypdAIWfl9iM9SdpcTgWNxgy4H1VlO3FMaKbe1cy4BultuQ+rVYYvJt9nxSYYB+HLI/YH9zOzEoqGcEpotly4kC5q74IMYKwPnq87m1iAIgpmJUOyCIOgXpA71Q+nnznm3BCUcRvn7q9Y7LZsJyFujPBDf2/NgMnHfSsZTMYn/4zqK6M/p/nLGWvwKNzBzc4nxlGrckzuewRVChtzBe7YHciyVPWY/SPssy9LMh+/vy/h3+v5y2vtXjR8xo9ER6F793pMp7vWQtALuvgLcDUOvYmbPAq+lnztXaxPJ9P/+OYUoK7flSlwL5NkFd7/QUqopUlTK/L9m9ng63pSKi4dulWkPyqUsr8upWOfdJnccBEEw0xOKXRAEjTCgynGzeRS5MH0PAa6QtEAxgqQtgKElVjehq1XFfJrBVIyE5F0PrJi+FywZ2c8rKwOpTrXZA8P92oG7Cbi0bGmY3GH4+G7ManSH7H4WKgmbfj85v3zgim62b2snSTPMCklaGC+3C8FN4lNx+bAUPuNVTDMI39OXt2KZ1a9wh+IzKO/JufX6ZnZ/yT3MU2MGZuP0XXQtMX1JraShueMhVMoLCmWWrpPtQy+r/+62/6K8WRmMAM4sKk3puifS1Xffhbnj0yR9boaLSEvgLhGu76ZcjVBtUKKszGuVaa1nqplygRnLF9yS6+Xp+FBJO5ZKHwRBMJMRil0QBI2wYO64THmoxZD0vWiNOKfhe53AO4gPSfpRWqq4nqQTcfP51Rx6f7rKUrR98c7pOQV3CC/kZDoy7beaU9L38D13GYtK+pyk7WC6UkAubTWOojIT8SXgAUl7SFpN0kaSzsH3wv2iRh6NkN3PWpL2TLKOkHQIPps3XWZJW0haL1m9/Gk6L+ASSSdIWkvS5yUdgBvJuKRguOMQYHI6/rGkKyVtIukzknYGHsA76Vfk0lwO3JaORwLjJB2WynYtSUcAf6d6/Q4FvlI8KWkH3Hfa33H3FXnyLiQ2SfG/jbu9eCEXdkxSYDNLorfie7UgWVKVlLe0mLneyLeFMoYXfh9NRZH+KnCbpK1Tm9gOt/y4IRXfi5jZWOD8XH7/kHRcWpb4WUn74T7ijk1+/1rNDLOEkr6Al+ejdPU7mS/Tk+U+Lgcki5V35sIWlbSQpKztNVwuiWxmd1h2Ig2q7ELlXXKOpPW7ea9BEAQdi/rQUFsQBB1IGikfjneoz8J9wYF3JHcDnkyWMaulHw58jYqi9A5usfH2EsfWSFoK+BvlRhSmATsk4yb5NOfiHblJuAGM/YBb8BmX3XAF62bcafrkXLo1cEMj2SDXe/jM1k34LFRmWn4K7jNvQ7wj/wO6OmPfAfehN0OnOl3jWipKQp6JwGbmDtx7jKSvApfmTr2DuyU4H5/JyZSeycCNwNeymUK5g/kjqmT9R2D34qxi6qxfSrlycz/w5eLMapqFvQ53mVDEgH2S0/F8mky2KXgdHwxcjbeHr+MzOY8Dm5jZ64W08wIv4uXwEb6f60lgA1yRvRfIBgM+xt0FTMaVmWNzcr6JPwPj8HZwdUo/AV9++KCZTU1K/0b4TKXwOt4EuC8p0UgajdfH/CVl8BywUXKQnr+POfGy3rQkDcDxZvaTKmENI2lXKr793sTL4iIq93MqXhcb5fzXIWkY8BgVS5cf4PX2P7yu/o7f98fAy/js7FMpbbfLJc3CL4/PHGfOzr8JXGlmk9OM38+pOKyfgu/7u8jMarrSGKYRNppaK4VnHW61a9otQhDMNEgaZ2Zl/30tI2bsgiCox4G4snQeFaUOfL/VI8C0/BK3EibQdfZrOK7oPFoWOXUSV8U7ZY/jjqDfxGd+1iwqdQXG4b6+9sM78+Nxn2l74QrU5HzkZFxkK+BBvOP3PL7naksz+wfeeX8XX1K5MT5L8yZd9+0sis8m3FDlfu7FZx9PBJ7FlZFX8fL8TKuUunSty3DH4U+k+3kKV4R3x50234PX5WnA9nlFzczG4ArJ33CF8H18hmV7M9u1bKlo8lW3Mq44vpru7TF8Nu8LZctlk+K1Ft7JfiBdZ2K67gZFpa7Aq7hy9Q28bv+L1/cR6XqvFxMkp+/b4z7W3sXraWsz+zg5lt8Sn0l8G2+rfwFWNrNb8Zmod/A9hZunuro9lWW2BHDeVK7ZbNRk4K+58LmBu3ClKJNpLO5I/DR8YGAa7tfvGGC1olKX0kzGXTB8O+X3XvrcgSvoLVPqSlgHr7OHgdfx9n8GsGpeqUtyvou3o5txxe813NXBZ8zsIeCEJPfteH0/lUvbSLn8FR88WCx37kLgPUkjcAuZeWfuc+DK6FuSuuNXMgiCoOOIGbsgCDqe3Izd7Wa2XnulCVpNbsbueTMb1V5pZg3yM3ZmNktZlhw9erSNHTu23WIEQTCTETN2QRAEQRAEQRAEQV1CsQuCIAiCIAiCIOhwQrELgmBmIHMjUGuvX9C5RP32PdNdc9TZQxsEQRD0E0KxC4KgY5E0RzJNn5kyXymZSJ/BX1zQeUiaXdIquKNpgPkkfSdZWg16AUmzSVoG2Cl3+ieSavkQDIIgCPoBYTwlCIKORdIkcjMLBYb0kk+vXkHSWrilv2bZyszurB+tc5D0CLBCleDlzezxKmGzNJLuAxZvMvm9JH9/JWxqZqXWX2cmwnhKEAS9QV8YT5mtNzMPgiDoTcxsZloilrlFaJYJ9aN0FmbWk/KYldkYGNhk2vfNbGIrhQmCIAj6hlDsgiAI+gHJefUr7ZYj6HzK/PkFQRAEMz+xxy4IgiAIgiAIgqDDCcUuCIIgCIIgCIKgwwnFLgiCIAiCIAiCoMMJxS4IgiAIgiAIgqDDCcUuCIIgCIIgCIKgwwnFLgiCIAiCIAiCoMMJxS4IgiAIgiAIgqDDCcUuCIIgCIIgCIKgwwnFLgiCIAiCIAiCoMMJxS4IgiAIgiAIgqDDCcUuCIIgCIIgCIKgwwnFLgiCIAiCIAiCoMMJxS4IgiAIgiAIgqDDkZm1W4YgCIIgCIJ+wTCNsNGs3W4x+gW32jXtFiEIZhokjTOz0b15jT6fsZO0rqQ/SZra19fuLpJGSPqBpAckndvH155X0n2S/ifp892Iv5KkcyU9K2mypEckHSRpzh7I0O/rqBNotC6DIChH0uySbpL0lqSt2y1P0BlImkvSHpLG9fV/eRAEQTtoSrGTNEDS5pLOkfQfSe9Kmibpv5KukbSnpKEp7qmSVpS0qaR/A7cDOwCzt/A+WoaknwPjgFOAVdogwvrAasDCwI61IkraBbgSOBmX9RfACsDRwI2NXrhT6qiD6HZdBkFQk5WBDYERwLfbLEvQAUg6FngG+APwmTaLEwRB0Cc0rNhJ+hLwCHANsCQwBlgd/8MdDZwJbAG8JOlmYC9gXuA2M/sscG4rBO8tzOxwYHngjTaJcAtwP/AKcGG1SJJWB84Cfm9m48zsHTM7GvhNirKmpOENXrsj6qiD6FZdBkF/RNJG7ZYhx0PAzcDbwP+1WZagB/RhuzoC+BTwTh9dLwiCoO3M1khkST/DZ4WmAl81s8sLUSYDVwFXSdoGuAhXHucxs/dTnLHArj0Rurcxs2mSngXmb8O1J9C90cWD8fobXzh/AHAf8KqZNfSH1kl11Ak0UJdB0K+QNAA4DVim3bKAv5OB/qRoBk2QVvIcB6za29dK/2fvS3oKH3wOgiCY6en2jJ2kA4CjAAG7lyh1XTCzK4Dd0s95c0FTGhWyTfR3OTdO35PyJ8250Mxu7kHe/f3egyDoXb4JLN1uIYKZjn3x1T19yfv1owRBEMwcdEuxkzQaOCb9vNHM/tSddGZ2Mb7XK6/YfdSQhO3j43YLUA1J8wBD088Pe+ESnVJHQRC0GEmjgJPaLEYwk5H6EYe24dLxfxYEwSxDd2fsTgQGpuPf1IpYwrHAPA2mCWqT3zsX/iqCIGgJkpYCbqLrYFwQ9Iik1F1HGOQKgiDoVeoqdpI+A9MdurwD/L3Ba9wK3FMjf0n6jqSHkrn+h+qZs5a0jqRLJb2SrHG+lEz+L1cjzXBJR0h6UNJ7kiZJ+peknRq5GUmnSLKSz3qN5FMj/+UkHS/p9WKekm6TZMBzudO35mS4rRUylMjUTB0tK+lkSU9Iel/Sq5Kul/T1GmlWS5ZWp+TObSrp3pTHU5L2LqTZSdI9qU5fkHSoJNW4xihJp0t6WtIUSW8muVq+f6dOXUrShpKukDQ+nRsg6YfJ0uxkSWMlrV8l78GSfi7pGUkfFtriBSnOHrXaqaSt67VjSQtKOkTS85J2TXIfLuk1SeMlfS4Xd64Ulj1j78vdcPxB0gx7tSQNlLSNpJsl3ZrOzSHpsFQ/70m6XVLN/TiStk11+Lr8ffC0pN9IqqqcqIl3SE+R9FVJ16ZnYbKkhyX9KLXxR7Kyl3RBsV4K+ZxUK7wQd6SkX6X835P0hvw9sqekgYW4W+L7c5fOnctfZ9VC/AUkHSvpsXQ/E1Pe27eguIr3saj8/f2ipF2rxNlJ7l5kSkHul1oox2ypvd0s6ZZ0bm1J90t6R9IRJWm2knRdap9T0zN7iqRP1LjOQEnflnSL/B31rvx9sLOkHeTvulEp7h2F+32ukNeVtcILcRt6LiStKenGdO/Fd8nSKc53gDuABVKyJQrxRhTybOodLenLkq5Oz9cUubuib9ZKEwRBMLPRnRm7bXLH48ysoWUNac/X1WVhkmbDrQWeBMwNzAGsBFwqaa0qaY7BN/X/Abd4tRpwN7ALcF/Zy1+unD6GuwTYAVgI+Fr6fZ6kMxu4pf2B69PxP4FNcSMrtzeQxwxI+qSka5Oc+1NuuGUDYBBd975smM4NSuEtpck62hl4EB+d3RQv7+/jJsv/kjoag3Px15X0T7xTuSswWM4pwBW4u4A58Ps+Wa78DJR0DnA2sAgwGFgMN+7zsypybYt3MMYBXwCWwtvSl4G/yY0D9Zh6dSlpO+A2fGZka2A2ScPwZctH43tQhuAb/q9TiVKEl8v2wM747MrqwA2FOGfjZX9uFVGvwsvu9JJ7WEDSRcCLwC+BxVPQCcCReCdtKeCQFH8e/Dkcg1trHYk/Xw8DewD/zndkJe0F3AtcjrdbpfB/4QaAhgFzAusCN6f8izIOkXQZ8Fu8HpfELdq+iO/lGSdpZEm6ht8hPUE+qHQ9cCpwcbrmkni9HIPPZKyQS7IbbrTk/ipZ/gRYFi+rWtfdGLdg/Cm8rSxM5f13JvDPQrleC8xHV3cC2btlkJk9kMt7Lfx5fR1/fhbHrRB+Dri4wXdqrXuYV9JfgKfxtrVolXgHAr/HDXMsjJff8bRwRYOkffFBtcvwNjtA0prA33BjIMOAMVmZShok6Xy8Pf8Sf17WBl4AfgA8KGkGlzqSFsHb40H4e3dJ4NPpOucAf8LfdRkb4G34+Sqi74A/i0/Vub+GngtJ6wD/wN+ny+Lvku8yoxXKs/CtA79Iv5+na7t6O5dnU+9oSb/F359P4ta5R+Jl9QdgzVr3HQRBMDPRHcVujdzx0y2+/tn4C3w+M1sCf5FPxZd97luMLOnH+J/MBmZ2o5m9bWaPAtvhlhznxDsVc+fSjMI70M8BXzOzx8xskpndAFyaou0pacNuyvwpvEx+CaxnZjeY2Ztm1tMOxAvAV/COeilm9pGZfUjXPQMfmdmH6dMbewkaraONgT8CV5nZd83smeSK4TJcCZ0CbAWcn0v2OD6AkDfIcyLwP2ABM1sM79j8N4UdiCsjE4CFzWxxXHm6I4UfoBlnI9bArbTuaGZnmdmrZvZfMzsMVwwAjkqdlZ5Sry6vM7MvUhkgmB0vjz8D85rZJ3CXIeAK648K97IRrjAfbGZ3pPK9L6WZPjtuZh+b2Wt4B3cGUvgreMepyERgn8I9ZIMII/HO5VQgG7T5GbAiPvhzcno2n8SNcEzBldVdc3ldgHfAHk2/58HbzVG4Fd0F8c4vzKhs5PPYEtjMzK5Oz/XTuOIJMAr4dT5BM++QniBpEK4wrQ2sa2bnm9lbqf2dABxWTGNmH5jZeKqY9U/hT+F1UO26K+GK++O4BeNHzezdZFRpPeA14PPAtdmzktrDh+T2F+feLR/m8h6Fu7s52MyOM7MXzewNMzsJfzbB36nf6lYh1eYt4FvUGLRKgyJjgP8zsz+lOh1vZj+h8a0DtbgIH0B5Mf2eA1dml8fd+ryFr1CZmMIz649fNrM703N6L7AZ8BL+zvpzGjzL7mVufJBwYeALZnZVSveimR1CySCMmU1NSvclZUKb2RQzewhvD6U0+VwcAzxuZoeY2Svpcyb+zOev39121dQ7Wj5Lug/wazPbP5XVm2b2W3xAcTBBEASzCN1R7PIjpG+2+Pp/NLMTMjP7ZnY33mEA+Gw+Ylqu8QvgotRZnY6ZfYz/oYLPXmyXC/5dOndkipfnjtzxwvWElfvw+ztwkJkdWpJf0+QUswfqRu5bGqmjwVQ6o8cVMzKzx6h0tL4uaYt0/jUze4Ous54XmNkvzezdXNrMoMNI4Nb0J/5mCn8bV7bBZxaXLVz+ROBRM/tHyT3mLYjuVRLeEPXq0pFjEoAAACAASURBVMwyS6aPpO/hwKGpMzMtxbkW990FXQdXoGK6Oz9qT7rmr0ou+V4dkScVT5jZtFQn+c7gCsCPzexlM9sRGGJmZ6WwT6fvtwr5vIsrF13kTUrYx8B/0qn5gN3M7PLcc3UGrrxDoQzSyP62wKVm9nBB/HFULOEtnEvT7DukJxyIK3UnJUW3yN9qpG243nKcjQ8YnGDWddDJzF6lYsTi88D36lynyFG40aYy34z5Z6nRfGfAnGnUfi+ugCtZi5WEHQNM66kcSZZXU9ndlU6tARxmZs+b2Rn44Nf6ZvaxpGWBHwJnWsWNTJbP5FweywFfygWfgM82jjGz10vEaHl76cFzsTqwoKTZC2muofpscy0afkdLWhFvy28DPy9Jdz4+q1wX+XaDsZLGftCaJhMEQdDndMePXf6l3VILjGZWtl/v2fRdXIr4VXw5xx5V1s0PzR2vBCBpMXxE/wN86VuRs/DR1TnxJVJVScsLTwK+ZWbX1YrbQyb3Yt4N02AdbY0rXZPxznUZZ5KW7+Edn2tyYdM7JmY2tiTt47nju0rC80uRpi8xkxuEWAuYIumVknT5Nr5SudhNUa8up6bvN8zskZLwp/Hlq0Xz4G+k759LGp+UwIwb8RmBVjE1d3yq5WaFCwrDKXjH79SSPN5N32Uj51n+482sy14oM/tI0vMp32IZfD9931TM0MzekfR5fBlnXjFt+B3SE9JARzarXW22pOXOk+X7Hkenn/+sEu1CfJBlKP4cntLNvIcAX8fd3vxPM25nzc+U99WzlD0PW0k6GDg2a6dm9oakMkWhJ2Rt9q78e6rwPOyED5weWWX5YH7mayXgprQEc5d0rs/aC80/F2/g7/sLJe1puSWVdF19UZcevKMPxPsxNyeFuQvpHfI4lf19VUmzjWcCDNOIMEoWBEFH0h3FLv+ynq+3BMmRvZznKJz/fPr+Bb4Eqzt5rJu+X89mQvKkP+K/1BNI0uH48q7Ne1mpg37sZiFHtTraNn2/XpwlyDCz5+XO35cE1pGkXNx6S0lrzVBAV39F+Y5A1naup6IQVKOVgxf16rLe/WaduKJCdAU+IzoCuEbSjcBR5ssyp9KCWceMNPuQ/axaNum5mP5sSJoT39uzK5XZtrIVAg2XQVo6mBl0KjWMYWYP4vs88zTzDukJn6diXfKJFuTXXbLn0KgyW2FmkyXdi88WLStpYfNlufVYHX+2xuKDZrVoWee40A6LYePTHt118BnrnST9ErjYfPn6rq2SI5E917XeFVlb+x71919nAx8b4//Jb1WZrestmn0uzsFny74GfEnSCcBpZjbRzI5qUoZuv6PTEtat0rmy2fCMTvhPDYIgaAndUeyeoNIxW6gXZcnIOgPFf/FsSdVs3eyAgI8mgu8LapZlgW+k40Mk3VymJM5iVKujzKhLvT/Sx3HFbgiunLxVO3qPydrOnA20nX6Lmb2Z9tldjBsX2BjYOHVu96sy29knSFoQNxjzVXzQZGfcSMgXW3iZ+agoeo082828Q3pC3iBKvWVyrSR7Do3aytXjVJYBLgJ0p0yyMhzWz56lr+FKyUb4nrcLgF9IOsTcn2pfk5WTGiinrL30ZVuB5p+LMfhs3j74M/kr4CBJxwPHmdmUGmmrydDIO3oZ3GAN9LOVLkEQBO2iO3vsbs0df07Vhk17n6wDN4MVsRpk9ze/pOE1Y1bnSSozIGtRsnk9mM5c6bvezG42C/wRldHq3qSZttOvScrbiriFxGwp2jrAPZJ6vLepGSTtglveWwhY3dyownO9cKn8e+uTDaTr63aQXz46tGqs1pM9hwOo7UM0vxqju4MrWRkuLWmumjH7kLRP98u4gpcZ5FkS+JOk8/MGSvqIZtpa1l76sq1Ak89Fmg3dFx/4zQxBDcf3ut2dBnl6U4a8O5PiUu0gCIJZku4odpdQ6Xx/gorhhr4m67xukvZ5VEXSF9Lhq9kp6rgCkLRutTAzOwc3wgKwe7IgFsxIZi1uuKTFa8TLBgeetZxVtF4kazsLJ/PkVVEVFw79EXNrd8fj1h/H4Ht/BgCnyl18TI/a27IkU/DnAn8zs13NbGKdJD3hTSrL4Or5tlolN6jTzDukJ+RnXj5dNVZ1mq23F3PHK9aIlz2H71NlSWsJWRkOpM5SzBaVYUOYW99dGbek+b90+lvA3lUT9Q5ZOW1TK5LcZcdq6WfWXkak/XaN0mx76dFzYWb3mdlmuMXke9PpVUj71RqUoZF3dH6/YSv3cwZBEHQsdRW7ZNUubzL84EYuIHekfEajgpXw7/Q9AtivxvU2wX0pQVfH6FWVMbnz1ZoOt/HlZZkhkeOTWf+gK3lDK5vWiJdtZL+sF2XJ8+/ccZnlNAAkzU/9/R1tR9KPlXMMbmbvmdmR+Izy+3iHPW+9Lr90uIsFuxK6M9hTlGde3P8ewHmNpm8UM/uAitW9zZIFwmr8gkqHuZl3SE/I76trxsrm9HorWh4sIilfb40+h1c1MMAylspS60OryZXOH9TNPHuEpFXlfuyA6eb1L8Q7+4+l0zv0hSw5sra2XBWDJBm74AMz0Lr20ugz3tRzIamLuw0zuwvfK5ed3zLtte0Ozbyjn8RdqYDv8as3QziwTngQBEHH091O3LFUnOFuK6lbf5Kps3Embn1y+unuCldY9nkJlT+uIyRtVRJ/niTrxQBm9h/ciS7AupJ+VJJmKN4ZLW4a7yJn6vhsh1sqHIj79Fm+u/fSAKpynCf/x90bS4yaraOzqXSiv1+2bDcZvlgFn106qxjcoJw1RcsOkiGNzCT+RnJHvEW5hFsGbKWyWa8uu/v8laWdobNoZuOotOP8Mrm3qYzmlylBeSuatTpi1eRdhsqet7J9uNkyq7KOVbNlkN3nbMC5koqGfJD0U+C5nCXPht8hPeSfVFY7fD+vjOeotQJiQu64S72lpYX52cp8vV1Oxefj7jVmYbJrF/0Y5n2LDc0dDzGzt6gsu1sBOKfKMsejKLdE3BSFd0nZ87Bj8X1j7golM+LRG8tGa7XdvCuI08rqXtISuEXSrDxvoKI0Hy6pbJlxd9rLQir4YUz1mPl/Kz7jzT4XG0rq8rynZ+2A9HMAvo86I2tX+TYlSYObeUcnQ1F/TcGzAydX2SqSnevrJa5BEAR9Trc6VWmEfDMqJuwvkLRPrTRyp7HnA5ebO2XNGJ6LU+/PdnpcM3uZyszhIOBySWdKWl/S6pL2xJW4y8zsf7k89qPyZ3mSpLMlfUnSGpL2wv1oPWHu4DlPtmY/25yNmU3AZ/beT+HXSxpJa8nvFai2LzDvc2/RKnF6QrN19BoVB8Ur452WItvjbhKOMnfCXO26w5iRfOe9TK689chi+nw7OFDS3yVtk0b7t8Ut1y1oZg2Z6a5DvbrM9iLWXP5UJe33JJUZJMmUqOl+n5Ifrcxq3H6SRqcO1ackXUDFeizAFyXNkZV/QSmo1tZeyB3/LJtBk7SypL8C2VKzRSUNKnTami2DM6nspfo8vrfwG6k+t5J0DfBTKh37nrxDmsLcX+GJ6efswM2SfippOUnLSjoI2LNGFg9RUciPl7SopAGS1sPba/55WE/SPJIGJuNO30tpF6Ti33E6aUndqrhT76I7gLxCuUmK/21g83TuQCoWaHfE91PtmMp+81TnW1Lu+L5Z6j1LK1Nxo5JnhuehBWRtteq7N+2BPT/9HA78Q9JxktaW9FlJ++HLFo/NjIyY2bNUBizmBe6S9F1JS0laQdKxVOqgjMwK7ADgd5LmS8/bV3AlO5vdmk/SipIWSNdt9rkYBJynrrPFUBlsfCAp1xlZu5ovewfh1qazZZTNvKMPpzKYuB1wkaSFwS3zShqDLxEFWEvSmlUGWIIgCGYKur3sKu2Z+SLuQNWA30q6S9LukpZOL9H5JX1G0pH4SNrJZvZX8Jkaua+anXPZ7id3jpqFj6JivhjgR4UO/pH4rFAm+574sqOxeEfvHnzpVV7u24HvUDGrvhtwC77043R8qc53s/iShkrahoqFsvUkrabKkqMnqCwbWQL/891E3V9yUpVUFvm9IHtL+kT2x5k6daPw/VQZ+6U//dmqjFY2cv0e15GZnYp3JA04UdKYdA8jUufwTNxp8PTOZrqvxXHT+Bn7Ku2NStcdSa6egL0kLZjSSr5EZ59C+CJZmZjZzSl91g7Wx2c27sdHgOelhcu1atVlup8VgK+ksAUl7aKkRKfO2IrAhil8hKQ9Cm1sNuA6SYdIWiZ16vfA9xNdmD13OY5P30viHcoP8Lb/El2XzB0AXA0sJV9ieUAubE9Ja6kwO5Y6e1enn6OAxyVNwGesLgQyP3vr4Xtpxqb2uiYVa5mrpOdoSCqDwUn5yGYoVkxKw+B0zSmp/J5J4SsDF+H1eSW+ZGzzosNlmniH9JCjqLiBGIp3oB/H3yOfpWub7UIq14vSz43xvXPT8Bme46mUOcCluBI5d0p7Na7cfYg/S6dKWlLSsNRJvgJfqVC29PhuKlYGL5b0X7xdXZ7yfhT3ZZfFWR2v5/txv5SrA1s1aBWxKvLZpvxS+t0lfVIzzhT+QtK56X09XNIG+HvmEWos72tAjjnk+7uy/dpLSto/PSdl7EVlNm52/Fn6J97GTgD+aGZ/LKTZh8oA6oLA74Hx6R6GUKNtmtndwB3p5874s/Y+vvd1Hyr73wDuxGfnMsW32efiy8A/JW2Q3kEr4ArtRGYctLiNiuJ2J74HclRmxbeZd7SZPYm3xWygYQfgJblLnddxa6/ZaqMFcDcxXWYzgyAIZiZUxd1Y7US+RGQ7fBZvCXz5lQEv439Kl+Gje/klPXsDJ1fJciW8c1m6F8HMuigskrbGZ4NG46OG/yEt+TSzUlP7klYHfoZ3JOdMaf4AnJlPI2kK5Y6UHzSzVSWdAvygJHyimTVtmUvS0rhFwTK+Z2a/l3QxPuNVjZ3MrJ4foloytLKO1sU7E1/A/5BfxjuMJ6e9GPm4e1Hd2uhquEJwYpXwb+Aj0VdUCd8odRiya62S7mN9fBbgRdws/9FpP2mPqVeXuPJzYJXwYXjnuGw27k0zm19uvKdYHlPxzt+pwLllfgTl1jIPwDs7j+CzBZcmZf0RvEN2spn9R9KidDXCked2M1uvkPcIvLO6Jd6JvQE4yMyeSx3sP+Mdrf3N7LoabTm7x+fwd0uRcWaWOd8mKf8H4527xYDXcOfOv0wzEaU08w5plqSA/BgfYFocX859Gt7mFweeTVG/ZGa3FdLOgS+B2wGfofsHcKiZPSBpV7yz/Xv8PTaD77PU3vfD3RosiJfP/cDpZnZDDZm3AE7CZ1Qvx91oTCzEWRJ/p26S8n4FL/tfmNmrtIBUdh9UCf61mR0kaVUqey4zPgKew/d7HWNmPXYhIOlmqhvhWrLMAmwalNsV2AMffCDJelIy9FJ2nTnxcv0WXq6PACeY2cXy2drMUvUM10xK5m+BLfD/5BuBQ8zsWfns1TeovCNmcHbeyHMh6W26KkmZ38TrgCOrlMfueJudHVc4f1Z0H9TMO1q+SuBwfHnycHw2/0Qzu1DSrfj78Vcls9OlDNMIGz3dVeasza12TbtFCIKZBkld+jC9co1mFLsgCIKg5ySluqpiFwR56il2QWsYPXq0jR3bNnegQRDMpPSFYtewBbwgCIIgCIIgCIKgfxGKXRAEQRAEQRAEQYcTil0QBEH7yJtg7w2T/MHMRbSXIAiCoCqh2AVBEPQxyZLrCLpaet1T0sJyX49BMJ3UXhbArTpn7C13adBK/59BEARBB9Mbzq1nSeTOkH/abHozm78FMizGjNbhGmFvM2uFY+aOpj/UZdB+5P7Y1moy+Z1mNoOj5xwbADcVzm2VPvviFik7Hkk74E6lm2U1M6tmnbVjZWmCb+NWnPPslT7b4C4+giAIglmcsIrZIpKvpaF1I1bBzF5pgQwDcV89zTIxObOepekPdRm0n2Q6fva6EcuZZmYT6kebuUl+CXviN+x1M/uofrTOkiXo34RVzCAIeoO+sIoZM3YtwswmAZPaLMNHuC+poAf0h7oM2k8oZj0nDRT1i8Gi/iRLEARBEPQGsccuCIIgCIIgCIKgwwnFLgiCIAiCIAiCoMMJxS4IgiAIgiAIgqDDCcUuCIIgCIIgCIKgwwnFLgiCIAiCIAiCoMMJxS4IgiAIgiAIgqDDCcUuCIIgCIIgCIKgwwnFLgiCIAiCIAiCoMMJxS4IgiAIgiAIgqDDCcUuCIIgCIIgCIKgwwnFLgiCIAiCIAiCoMMJxS4IgiAIgiAIgqDDCcUuCIIgCIIgCIKgwwnFLgiCIAiCIAiCoMORmbVbhiAIgiAIgn7BMI2w0azdrbi32jW9LE0QBDMLksaZ2ejevEbM2AVB0DIkDZC0maSrJT3d4nw3l3RNK/OdFZG0tKSnJI2XtHS75QmCIAiCoDWEYhcE/QRJq0k6T9ILkqZKel3SvyQdLmlVSYtLurbdclZD0s7Aw8C1wBbAwBbnew2weavynYXZGlgaWArYqs2yBEEQBEHQIkKxC4J+gKR9gXuBJYGdgYWB1YEzgO2A+4HngYXaJWM3uARYGbitF/JdCbilxfnOqlwJPA08A/y1zbIEQUNI2qjdMgRBEPRXQrELgjYjaVvgN8C/gfXN7DYze8vMXjCzPwKjgT+1VchuYGbvm9lHwLheyPdj4L5W5jurYmbjzWxpM1vKzMa3W54g6C6ShgLHtVuOIAiC/koodkHQfn6evk81sw+KgWY2BdgNeLRPpWqe93sp3ym9lG8QBJ3BvsCIdgsRBEHQXwnFLgjaiKR5gBXSz6omas1sKnB8nwjVcz7qsHyDIOjnSBoNHNpuOYIgCPozodgFQXtR7vjAtNSoGpcDH/eyPEEQBP2KpNRdB8zeblmCIAj6M6HYBUEbMbMJQLbPaWXgDkmfrhL3HeCpsjBJs0v6kaS7JU2UNFnSg5IOkjRHIa4kbSjpCknj07kBkn4o6T8p7VhJ69eSXdKnJJ0l6flkxfM5ST+njzpfkgZJOlDSk0nmuyWtUyfN6pLOlfSMpCmSXpZ0maQNa6T5YoozPnduJ0mPSHpf0sOSts+FDZS0j6SHUvhTkr5bR66Vk0XUFyVNk/SqpEslrdFImXSHVAZnSHpX0qhC2EBJ20i6WdKt6dwckg6T9LSk9yTdLmnVKnnPLelkSS9J+kiS5T5HpThHFc5bXg5JP64VnuKMknS0pNckrSdpsKTfSpog6YG8GwdJ80s6QdLjqT4mSXpC0kmSFi65hzkk7SzpHknnpnPzpDxeSuV2bVGmQh4DJO2eyuqt1NYek3SEpCE10m0l6Tq5RdypqZ2eIukT1dL0hFTf35Z0i6Q3072NTfe/g9xC76gU945CnTxXyOvKWuGFuMumdvJEqpNXJV0v6eslcb8D3AEskE4tUbjOiEL8UZJOT+11Srqv6xVGV4IgmBUws/jEJz5t/OBWMC33mQYcAwzvZvpFcKuZZwHLA8OBrwAvpPzGZnnhFjZvz13rOWAYcBMwCXg5FzYFWKbKNXfE99LdgCukw4AtccudU7K8W1xOYwoy3wK8A7yYk3kS8Mkq6Q8CPgB+AYwE5gW+Dbyd0p4GKBd/a+CBQlkNBq4A3iuU1ce464AhqUympPL/OBfnW1Xk2hu3UPl1YD7gk8DpKc0HwM4tKr81cIul+bY2Khe+F26gJgu7DfhEOjcReC0X9gYwTyH/wakd3gWsCswNrAPck9IcleINTOV/QxU5BgKLp3LuEo67abgGX5abhX0JuKxwX79N8UcBL6X62A5/NlYB/pXiPQPMlbv2ofjgSZbPucCnU7w3gAm5sCeA2UrKeX7gn8AjwHrAXLhV10dSunHAsEKaQcD5Kd1aSc58fb0OrNLi52kR3BLvU/j7YjiwGPCrQvlmZT841etzlDzfwBz4u+DJsvDC++593OLvJ9N1vwr8N6W7Ehiciz8AmA3fi5w9h7Nln0Le26b63gO3IDwSf96z+/lZd8pmKHPbemzerU8QBEF3AcZaC9/jZZ9ezTw+8YlP9z7A0YWOadaZ+xEwe410g3BrmmeVhK2cy+usdG5o+r4unX85daT2yK6D+4rL0p1Sku8GwIf4KHqxY7Usrpj2pmL3P+Aq3KDMoBS2TU7m40vSfjeF/bokbH0qCtixufOfABYE7k5h/wXOA74HzJHirIUrPYYrMFcAB5M67sCiwGMp/JGSa2+NK6PLlYRdntJNpYqy2mD5zY4v/T2UQqc9axupE50pIA/iCv+2wIAU5/u5tAcU8t8znf9M4fxQvDN+VOH85mVy5MK3KIbjStII4Ke5sAvS76VwZfE9YNMU/9IU55JC3kvn0u9YkHUI8CYV5fYW3FptFufYXNqvlZTxXbgCuEghbLdcumJZnIT7ahxSOD8nlYGLxylRJJtsC3PjCtiLwAIl4adUqxvcKmUtxe34auHAxinszyVhy+MKnwF/qfH8V7vuGrgCv25J2G9y97NOvfIJxS4Igt6APlDsYilmEPQDzOxg4Ft4hzJjfrzD95ikr1ZJugveofltSZ4P4cohwE6S5jKzSen3I+l7OHComZ1lZtNSumuBh1J4l6WAkgYD/4fPqvzEzD4sXPNJfEalNxkGHGZm51iyImpmV1Bxs/DZgswLU+lsnlDMzMxuoeJOYn9JK6fzL5vZa/jsDnin/0gzO93cUilmdic+05Jd91gzO9rM3k3hL1GpmxUkzZmTayBevzea2RMl93lz+p4dn1nsEWY2Lf2xPFAlfJK5W4n/pFPzAbuZ2eXpPPgsy4R0XFwmunr6XqyYL96xLvJeHZEnFU+Y2Xtm9jZwde70cDM71syeNrNN8MGL61NYtqz5rUI+4/GZ2i7ypjJ4H/fzBzAPsE1qIxnHUDF0VCyDvYE1gd+b2f8KYXfm0k1fAippWeCHwJnp2nk5J+OKIsBy+OxkKzgBWAYYY2avl4T/rUbahusNurw7oMRlgZk9RqWdfF3SFnWuU+RE4FEz+0dJ2M25472qyPedtAx17AdMa/DSQRAE/YNQ7IKgn2BmF+KdtzPoagHyk8Clkq6SW9HMs0v6vknSK8UP3jkHVw6Wy6Wbmr7fMLNHmJGsY1s0Lf4tYImU7i7KKcuvlbxhZg+WnH82fc9fOL8LrpQ9kRS1Ms5I3wOAHxTCso7sW2b2NDPyeHZQpUyezx3ny3M9vCw3rlJ3R+firlRF7maYXCc8axvjk2I6HXM/hdn9FNvGG+n7VEmfK4Rls4+tYmru+Hf5gKS8ZhyPK0bnlOTxbvoeXCP/+81sYiH/CVSUwmIZfD9931TMMA16rIsrcQflgnbC292RVdrBlrm4PW4Hkhah8t64qkq0d3p6nRK2xpdGTqa6r8szc8c/7G7GkpbCZ88/XaUML8hFLy1DMzvTzEab2ehBYaMlCIIOZbZ2CxAEQQUzexPYS9LvgF/jy9EytgT+IekLZvZumvHJZgzWwPdj1SI/G1jPdUDWsSt2er+Wvp+skbZdljszhWWOwvlt0/erNdLehXfmBwNfLITVK6vSGYoc+VmYfI/x8+n7/+iqxJUxtU54I9Srn2bbRrYkciRwl6S/AL8ys4eSgnhYw5JWJ38PH1aLZGZnA2dnv9PAyM74AMUi6XTZAGd3ymAecmUgaSS+HBR8n1eZPHfgS5jzZO3ge/j+11q8Wye8O2yM//e/VWW2rrfInsPXC8r3dMzseUnPAksC60hStbgFsjK8nopyXY2q7SUIgqDTCcUuCPohZvYfYEtJ6+EzEtko84q4MYAf07Vj+YGZvdIHon0mfdeb9WkHWQew2FHPLCRWVWjM7ANJz+D7fBapFq/FZMvxBvdR3fUqZva4pK/gBkcWArYHtpd0NbC/mZVadO0LkmXHg/D9lOfgAyb34DOmrWJk7nhQA+mydqA+ageZ38x6SypbTd3nMPE4rtgNwWdE36odHaiU4Zwzw7MUBEHQLLEUMwjajKQLqoWZ2W3AaODPudO7ShpA187jKr0j3QzMm76LS9D6M3Ol7+ISzSLZ8rrudCRbQVZ/fVV3vY6Z3YAv+f0VlZnMLYGHJW3V1/IktwM/wZWFt4GV0h7IWrO3zZL/P/1kA+n6uh1kz24tn5m9QfYczlczVuU5/Ijuz1DOdM9SEARBM4RiFwTt57NJUSslGTXZlcoesrlxn04TqIx+b1PrAnKfc/PWitNNsmV4y0vqlBn/F9P3spJqbZ7JnMX31cxStidtjbSMryqS1uoDeVqCmU00s5/hrgZOxtvoYOCiwn22cs9dNU7ErViebGYHmVkrl7QWySuLNX2mSVo798xn7aDeMzxE0mo9kC8jm6kbkfbbNUqz9ZY9h8MlLV4jXvYcPls0zlSDrAwXlrRmrYid9CwFQRA0Sih2QdB+5mXGfV1dSFYYr0w/PwYmpk5qZr1yZ+WcMpdwGK153rPrzYWbq6/FwBZcrxX8PX0PprZVwcwB8mW9K850/p2+B+Km3EuRtDyVvY39FrkD7+mKgpm9aWb74D7SPsZN9+cNgeRND9azVtFw25W0ArBP+nleo+kbxcyexX39gT+PpQMpaXBh/5yl0awdLCfpmzUusQuuLPeUvAXW7ZpIn9Vbo3X299zxpjXSNfMc/jt3/PNqkSTNT/09eEEQBB1LKHZB0D84PBlDqcWw9H1LZm4fuDB9DwGukLRAMVEyGz7UzN7Ine7us6/C7/yS0OMkzV0jTauXehVl6W6806jMbBYtXnoC74SPwg3M/LkY3M3rNirb36i4o9hD0gwdzmQi/gy6WvVrpQxl99Zs2xgIfKMYKbnPyDr1c+WCJuSOly3JP9/5n7MkPKOavCvmjhfKB0gSlT3mZc9ds2WQPY8jgDOLM/HpuifijsiLaQBOK7EoiqQlcCuR1xfDmuAGKs/D4ZLKlo2uXnIuI6u3hYrPv6ShuFN6mLHOzqYyW/j9VBZdSO/AVXBjQWcVgrPZu6G5+JI0OFnJfTidtbTpIgAAIABJREFU3kjSMSV5C/fP11cDN0EQBH1OKHZB0D9YDzgjdeRnIHW+tscd8P40F3QaFXP7KwIPSfqRpNGS1pN0Iq4UHERXsn0uQ+rINbzw+xwq7gyWAW7Nloel/Uzb4b68wJd67SDpi7WWmjZAJstcNWMVZE7+/DL/WFtW2eu1F97B/7GZFffYZfkNo5zpVjgllcmWr9PpeSSfZfl6OVXSJZI2lbRamr25F3jazKqZh2+G/P7IYv1C820DYIykT5WcH4Qv4cvP2oyn0tEfI2m51FFfXdL1dJ2dWk/SXJIymfKyLVpFvhdyx8fI/RkiaW3gViozQ4tKGirpyFz8ZsvgaCAz3vFV4DZJW6f63A63iLkh/twCYGZjqfhCHI5bvj0uLdf8rKT98HZwbG5Ap2nSzGI2UDAvbsH0u5KWkrSCpGOpPRufuRoZAPxO0nySBiXDObfh7yiA+SStmA02JVcjB6awlSl3Z7A9vhf2qORrME+mUM6X3m8CjqRiWGo/KgrrgZL+LmkbSatK2ha3OLqgmV1e496CIAg6m2Y9m8cnPvFpzQffH2Lp8yjwHdxs+lB8JuOH+BKvCcAmJemXwv3OWclnKu5gOYs7ELeK92Iuzi7AXCl8EK4gvpQL3wO3NpflsWTJ9V7CZ7v+jTvdzs4/g5txH9iD8hmEd96ez+W7OzAsF/7pwj3tBgzJ5TEAt9ZouPuBvfGO/QLAT/DO6N6F686W8n2hcN25cuGfxB2YZ+EH451lpWuOpOLDzXDXBvMVrnNklbozvKM8Z7NlV7iO0v1en8v/vHQum8FaEzd6YqlMNsnKEVdQv4AbtDDcvcbmuFVPcvX+Jr7cbQm8k35IOn9UiUxHFe73Q9xoxg/wwY78+ctS21sYOD0X9iA+yzN7Ie+BwH25eB/hhjlexpfk3psLeyPd2+B0Tx/n7uWzuXscQmVpqaX81sxfGzd29HqV+nwWWLqkHOYErqvRDo5r8TtnbmBslWudnMon+z2qJP0/S+ptQirDMbnzE/FZskGFOv841ccY4BP4YMO38bZ3NG4htHjNT6c02XvtFeC8Qpw9kixl9/UIrtjVLZ+hzG3rsXm3PkEQBN0FGGst7kMWP72aeXziE5/6H3xp1NzAavgMzk2p0zI1dXQeTJ2dRWrkMQxXEB7DO+Rv4ArFaoV4x9ToPA7FFYmysDcK+cyN+9l7Jsk5HjgC7xiPSZ2onYDZWlA+x9eR+Y4qYc+V5LUV3oF+DVfwnsSdIq9YErdWWY2gqwJb/KyJu6SoFr504VpfBK7GFYkpqfx+SkFZ6WE5blhDnk2Ai2vVPfBclfCxKbysPCYD/wC2qiLTgNRuXsJn724FvpTC1kvl8WtgiXRu7Rr3cG5J/iNxhfAtXNk6E5g/hX0LVzzuBT6Xzt1d5x6rXfvSwnUXBk7FBxum4gMhRwMjatTPAHzg4F/4cz8JV6C+2kvvnTmBX+IDJu+nctghV/a1FLt58VnGt3CF7k/AkilsDL6Pbx9geJVrrwtciivZU/D3yEXA5+vIvDvw31SXx1HyfOBK/p9wYzbZu+lXpIGg7nxCsQuCoDegDxQ7+XWCIAiCIAgg+c+8Nf1c0syea580fc/o0aNt7Nix7RYjCIKZDEnjzGx0b14j9tgFQRAEQRAEQRB0OKHYBUEQBEEQBEEQdDih2AVBEARBkCfvqqSeFdogCIKgnxCKXRAEQRAEmV+4BXCrshl7J5cGrfTnGARBEPQCs9WPEgRB0DyS1gL+2oMstjKzO1slT6ci6WRKHIB3kxfM7DOtlCdoD5LuAxZvMvmfzKzMf1zGt4E/FM7tlT7bAFc2ed0gCIKgDwirmEEQ9CqSZsfNozfLBDOb1ip5OhVJc1PfYXY1PjKz11spT9Ae0ozawCaTv29mE1spz8xIWMUMgqA36AurmDFjFwRBr5KUslfaLUenkzrk0SmfxQkFPQiCIKhG7LELgiAIgiAIgiDocEKxC4IgCIIgCIIg6HBCsQuCIAiCIAiCIOhwQrELgiAIgiAIgiDocEKxC4IgCIIgCIIg6HBCsQuCIAiCIAiCIOhwQrELgiAIgiAIgiDocEKxC4IgCIIgCIIg6HBCsQuCIAiCIAiCIOhwQrELgiAIgiAIgiDocEKxC4IgCIIgCIIg6HBCsQuCIAiCIAiCIOhwQrELgiAIgiAIgiDocGRm7ZYhCIIgCIKgXzBMI2w0a3cr7q12TS9LEwTBzIKkcWY2ujevETN2QRAEbUDSAEmbSbpa0tPtlicIOhFJgyRtK+n6eI6CIJjVCcUuCIKOQ9Jqks6T9IKkqZJel/QvSYdLWlXS4pKubbec1ZC0M/AwcC2wBTCwvRIFQechaUfgduAyYBPiOQqCYBYnFLsgCDoKSfsC9wJLAjsDCwOrA2cA2wH3A88DC7VLxm5wCbAycFub5ZilkLRRu2UIWoeZXWRmXwCuarcsQRAE/YFQ7IIg6BgkbQv85v/ZO++4O4qqj39/IYX0AKGDhi4iPTRpoUovguArAhFBKQpYXwVUQASkKCogIAIKCAIC8goiIAREaQEEQXov0kNIIAkYzvvHmc0z2ey9T7vPc5+bnO/ns5/dOzNn9uzs7N45OzNngLuBzc1sgplNMrPnzew3wFjgkqYq2QHMbJqZzQTubbYu8wqS+gFnNluPoEd4rNkKBEEQ9AXCsAuCoJU4Nu3PMLMPypFmNh34AvBwr2rVdaY1W4F5iL2A5ZutRNAjTG+2AkEQBH2BMOyCIGgJJC0ArJJ+1nTna2YzgFN6RanuM7PZCswLSBoDnNZkNYKe48NmKxAEQdAXCMMuCIJWQdnx/0oaViftlURjLwAkLQfcCCzYbF2CIAiCoCcJwy4IgpbAzN4Cnkw/VwNul/TxGmnfAZ6oipM0UNJhku6UNFnSe5IekPQdSfOX0krSlpKukvRkCusn6auS/p1kJ0ravJ7ukj4m6VxJzyUvns9KOhYY2Mli6DCSdpJ0vaQ3JH0g6VVJN0r6TDtyi0o6RdKjkqalMvqrpO3ryAyS9DVJd0l6J8ndL+lQSZWeClM57ivpFklvSZqeznm8pFEV6TeRdKmkx9PvUZJOlfRiug93S9qzJLMjcB/ZEExJlm1rZOGjU37FdU+V9Jik0yQtVqHP/JL2Sdd8QQpbINNpiqRrU29hrXLrJ2k/SbdKmpTK4BFJP5A0uI7czpKuS95gZ0h6WtLpkpaoJdMVkvfZcyRNza75e5KeSmX0L0kHS1IN+TGSzk7pZ6S69C9Jx0kaXue8XSqXinx2L93vYju604URBEHQCphZbLHFFltLbLgXTMu294ETgREdlF8c95p5LrAyMALYCXg+5TexyAv3sHlrdq5ngeF4789U4OUsbjqwQo1zfg6fS3c9bpAOB3bEPXdOL/JucDkdl/L9M+49dDRwMPDfFH5wDbmtgbeBs5LccODz+JBRA46skFkG+DdwG+6ddCiwPvCfJHMd0L8kU5TjH4E1030YBzyYZJ4GlkppPwM8ULoPC6VzfpjpVmzfz87TD+gP7JfF9y+2LN0Y4MV0P/ZI+qwO/CPTZ2iW/ij8w0GR5wXAx1O6N4C3srjHytef8hgN/A14KF37UGDV9NtwxzrDSzIDgAuT3IZJz3Vw76oGvA6s3oD6szlwe6lc50/32LJ6VGznVeSxdqpLbwJbJV03BB5NMncA/RpULkdT8Ryl+39EinsGGA8sWXU/8m0YI20c23doC4Ig6CjAROvpdlJPnyC22GKLrZEbcEKpUVk0aA8DBtaRG4B70zy3Im61LK9zU9iwtL8uhb8MXA3sX5wH2D6TO70i3y1SI/j2cmMSWBE3TBtq2KVGcKHTKqW4i1L44xVy6+CGzaUVcYWB+yGZAQuMAp5K29CSzDGZHvuX4q4Cbig37PGlK95LMjelsBElo+LZJH94avTPD3wF+CDTceNSvuMLXWqU2RUp/vJS+PLZNXwuCx8GDMaNFsMNq5txT61FmpMy2d1L+Q7EDZu3gMVLcV/I5I4rxZ2Gr384uBQ+BHghyTxarmtdqEPDk44XZ7qcAfwIH9LaH/gsMCWL36uUx8QUfnIpfMtM5pMNKpejqXiOgAXwZ/5aYGRHrz8MuyAIegJ6wbCLoZhBELQUZvZdvBfpzSx4NN7ofUTSbjVE98WNl59V5PkgbhwC7C1pqJlNTb8fSvsRwFFmdq6ZvZ/krsV7mUh5z0LSIODX+KLJ3zKz/5bO+Tjwp3YutyusnB1PKsXdk/ZL54FpKN2vgEHA9yvyvL1IyuzrAx4LLAv82MzerSEDbrAV59oS2AX4hZnNNg/SzF6hrby3kLSCmb1j7u20yG9J4DdmdpqZvWtm083sdLxnptDxyIprqEcxpHe28jKzJ/FeJ8jKzMymmtk03KAFNyB2NbObM/ETaXPyM1vdwA3R9YGzzOw/pbi/Z3J5ua0IfBU4J5071/M93CACWAnYrPoyO4aZTUl1/NYs+N9mdqSZvWVm/zWzS/Ge0IKjStlUliltdRBK9ZAulEstJC2N9/w9AOxkZpPbSf8l+bDqiR/wfnvZB0EQ9EnCsAuCoOUws4vxBuzZzO5ZclngCknXyL1o5uyb9jdKeqW84cP7wHsNVsrkZqT9G2b2EHNSNO7L88I+D3w0yd1BNVX5dZcb8WGf55jZy6W4KWk/qBS+IT708PlkcJY5Bu8x2cPMbodZhusXs3POhpndiPdoHgj8NIsq7sN5Ne7DGlnaVbPjorX9kpldXaHjacBr6XjzzszFwr2o3gGcXxFXq8ygrW7cXzYczOeEFkZhuW4cnPZV5fY4sAluxH0ni9ob/88+pka57ZilzcutO8yycMzsjApdLwf+mX5+TO6opuBHuGF1eUlsSnZcLtOulMscSFoTuBPvgT3AfM3IupjZOWY21szGDui5qa9BEAQ9Sv9mKxAEQdAVzOxN4EBJPwd+DOyQRe8I3Cbpk2Y2JTnwKHpN1sGH7dUj7w1sr1H4TtqXG6m7p32VoVTQcM+dZjYJ2DYPk7QRPoR0uyKoJLZp2r9YI8/38XlkOWPxIYD15K6rCN4g7XfG56TV4+124vNzfSDpOnzY5QBgOTpoOJvZecB5xe/0UWAf3DhfPAVXfQjtSN1YgKxuSFoy6Qa1y+12Zu/xhLZyO4jZe9KqmNJOfCO5mjZjfGXShw4z+xFu3AEgaXG8h2+PTLZfFt/VcpkNSdsBvwd+aWbHdOZCgiAIWp3osQuCoKUxs3+b2Y748LN/ZVGfAH6YjvPG9Qdm9ko7W3uGX0dYK+3fa0BenSZ5FtxL0v3AN/D5Uv9bI/mSaT+gE6dYMjvujFwxjO7DDtyHzi48nS9MP7KTsoUXx7OAu3CjdQfgpc7m0w7dLTd1oNzKw2J7krplLmlVSZfgcyrfxnvdquhqueT0x4daDwPGS/poF/MJgiBoScKwC4KgZZB0Ua04M5uA9yL9PgseL6kfszcUV+8Z7eagWDdtDtf9PU2aj3UnPl/uS2a2axoaaTVEiv+CZTtxmvz/ozNyxb3oifswucZxXZIR/C3c8cjbwKpmdoKZvdpoBemb5dYdKstcvqzIKbgTlbtxb51n1Jnr1tVyyfkv8Gnca+3CwDWShnYxryAIgpYjDLsgCFqJdZOhVkkaMjged20O3oOwMO5lrxj2uGu9E8jXnGvEYtbFEM2VJfXasHdJK+Bu+lcCNjOze9oRASgMmIXS/KR6+Rc9LrnRs1UHZcCXA4D278Pikpapl6aColf2v7j3zI7yU9yL5S/M7DtmNqM9gW7QmXLbKKvvHS23we3dwwaTD0GetdYj/oHlG8DXzeynZUc5FXS1XGbDzP6Fz+M03NvthbXW2QuCIJjbCMMuCIJWYkHa5oNVkobvFc41PgQmp4Z64b1yH0nLVwo736Mx78bifENxJyL1qFzEu4scjzuCuanCeUot7sqOD6+VSNIWtM1VnEjbHLNDJFUOn5O0ED6/r+DutN86zf2rxWFU93bWuzdFT89dmVfTukhaBTg0/fxtR2S6g5k9Q5uTl31qfUSQNBD4RmYQFeW2kqS96pxiX3xdvoZS54NKUeYvmdmj6Xhb3PMpdLBMu1EuVXldiXtsBTeEj62VNgiCYG4iDLsgCFqN7ydnKPUYnvY3Z/O0Lk77wcBVkhYuC0naAV+/7o0suKPvyXKvQD4k9GRJVXO+CplhHTxHR/hE2i9aETfL+CqV4U3AK+l4b0lz9ApJWgw4nVSOZjYFuCZFL4f3eJVlBuBOSXIvlsV9EHBplZEtaW18Tbj7K65hgTo9MJ9K+9NL4bOWmpA0LDseTFt5QanM0nmK3taqOtfVulGUwSjgnLLRlM77U9yrZFkG4ExJ681xEp9T9lV8YfpGU6sXu6rM65Vp/gGgXKZdKReYs3zBPblemY6PkvS5Su2DIAjmIsKwC4Kg1RgHnJ3c7c+BpGWBPfHFtr+dRZ2Jz6ECb3g+KOkwSWMljZP0U3wB77Ir9WIZhPbc548o/T6fNq+MKwC3FEPk0pyuPfB1uwBGSfqspE3rDTXtIM+n/YaSDkjnGyXpCLw3r2ApSTtIGpeGsBZlJeBySadK2lDSBpK+iTvJuDytNVdwBG3OYQ6XdLWkbSStJWkf3BX+KHxB8YIr8QW9wR1m3Cvpe5LWS+f7AfBXaru0HwbsVA6U9Fl87bS/ApeVot/KjrdJ6b+I96Q+n8WdmAzYwpPoLfhQXvDyGiYp97TY1bpxAm2G9G7ABEm7SFoz1Yvb8YW8zywEzGwicGGW322STk7DEteV9HV8jbiTuuB0piPM0Uso6ZN4eT7M7OtD5mX6C0kjUp3fDl+PrmApSYtKKupep8slUfTsFh90fCV6770snvnzJW3ewWsNgiBoTayHV0CPLbbYYmvUhs8zsrQ9DHwJ7y0aBhQLOL+GN+S3qZBfDnfHbhXbDHyR6SLtfMAqwAtZmn2BoSl+AG4gvpjF7w8MyfJYpuJ8L+LLKdyNr71WhD+Nu7Kfr5tltFvpfJPx5R3OAz6Thb+LG1n9Mtmja5SN4csd9Ks433a4cVclcx8wukJmYdwIqZL5EPhKhUyh2zR8Mfn98d6gBVI9eBe4F1i4QnbBFG94791LuNHWL93n+7Lzz8QdqLyMe1rN9XwD+CQ+r2z7pKul+7kuMCidbzBufBbxb+MLbw/MdBqbrqOqDJ4Blq+4jiHAdXXu0ckNft7Gl67928BSuAH1GfxZexpYriQ3nNmfi/fxOaeP4fPeiuueiT9fK3SlXNJ9WAM3JIs0nyM9g+n+HpfFTcPXyhtV77qHMdLGsX2HtiAIgo4CTLQebidFj10QBK3ERPzr/Fp478Vn8B6AN/HG+f7Ar4FVzOz6srCZPYU3BI/Fv+TPSLJXAeubWd6z9CO8x22pLOwCYGoazncjvrxC7qb9V2S9FebzhtbChyk+gzdwpwM/BzbGG/wP42umrWhmv7QOLKZcDzP7A75w+GPpXE8AnzWz/YA/4fPp3sZ7Pfa0bK6SmR2NO664AW+IT8PLd08zG28V85rM16pbDb8fr6ZrfATvzfukzT6stZB5HV8U/Rt4r9403AC9AdjCzMpDKXNexY2m/8GddbyEG8Q/SOd7veJ8b+G9uE/ha7xdD+xiZh+m8t4RN3Lfxj8KXAasZma34D1R7+B1b3sz+we+jtyfaBsCuGAq16I36j3gj1n8SHwB9N9lOk3EFxI/EzeC3seNpBOBNc3syYrreA9fguGLKb9303Y7sLuZfatOuXWXjfF79i/c8DoWOBtYIz1XuZ5T8Hp0E+6h8jXgLGAtM3sQODXpfSt+v5/IZDtTLn8E7geWzsIuBt6VNAr3kHlkFjc/cAYwSdIaBEEQzGXIDcggCIIg6LtIOho33p4zszHN1WbeQNJ4fEgxZjbPeJYcO3asTZw4sdlqBEEwlyHpXjMb25PniB67IAiCIAiCIAiCFicMuyAIgiAIgiAIghYnDLsgCIKgFRia9o1cGiKoT1Hmsy0TEQRBEPRNwrALgiAI+iySBkpaHV9oGmAhSV+SVF5CIGgQkvpLWgHYOwv+lqR6awgGQRAETSacpwRBEPQRJG2Ie/rrKjub2d/bT9Y6SHoIX3aiipXN7NEacfM0ku4DPtJF8XtI6/1VsG2Vx9m5iXCeEgRBT9AbzlP692TmQRAEQae4B18br6u81X6S1sLMulMe8zKfwtfo6wrTzGxyI5UJgiAIep4w7IIgCPoIZvY+8Eqz9Qhan6r1/IIgCIK5m5hjFwRBEARBEARB0OKEYRcEQRAEQRAEQdDihGEXBEEQBEEQBEHQ4oRhFwRBEARBEARB0OKEYRcEQRAEQRAEQdDihGEXBEEQBEEQBEHQ4oRhFwRBEARBEARB0OKEYRcEQRAEQRAEQdDihGEXBEEQBEEQBEHQ4oRhFwRBEARBEARB0OKEYRcEQRAEQRAEQdDihGEXBEEQBEEQBEHQ4oRhFwRBEARBEARB0OKEYRcEQRAEQRAEQdDihGEXBPMQkpaU9ANJL0gaXyPN8ZKmSDqhl9XrNJJ2kfSWpJskDWy2Pt1F0gqSTpH0hqRxzdanFpI2kHSepKmSxjRbn6BzzG3vgUbz2L1PNluFIAiCLhGGXQsjaStJv5L0lCSr2D6Q9I6kpyVdL+nrkoY1W++g95E0UtLFwBPA0cBSdZIfCgwDvtILqnWX/YEFgC2AVZusS5eRtKKkPwGPAd8AFmqySpVIWlvS1cBtwBeAoU1WKegEc/F7IAiCICAMu5bGzG40swOAVYDnsqivAhsAqwG7ATcDWwOnAk9LWrcn9JG0jqRRPZF30D3MbDLeEN+8A8l/DrwL/KJHleogkraqE30uMAn4K/Cv3tGoR3gW2BXYr8l61MXM7jWzXYBvN1OPdupEUINWfg90h6gvQRDMK4RhNxdgZtOBiVnQFWZ2p5k9koy//YG9U9zCwE2SVugBVU4EwrDro5jZ+8CDHUh3hJkNM7MjekGtuqQe5pNrxZvZ1Wa2oJltma6vJTGz983sA+CeZuvSQR5r1okl9QPObNb5W51WfA90hzRU+DtNViMIgqBXCMNu7uHdepFmdjFwVfo5HB+G0zAkbUzHvgIHzWVGsxXoJF9j3vpYML3ZCnSQZuq5F7B8E88/N9Bq74Hu8D1gvmYrEQRB0BuEYTf3YB1Ic1V2vGWjTixpAeD8RuUX9BxmNrPZOnQUSWOBo5qtRy/TKvfnw2acNPW+nNaMc89NtNJ7oDtI2ok+Prw5CIKgkYRhN2/xSna8QCMylLQwcD2wXCPyCwKYZdRdB7S8p8ugMUhaDrgRWLDZugR9H0nbApc2W48gCILeJAy7eYsls+Pn6yWUtLOk6yS9LmlG8qx5uqQlsjTr4nM1cmcsz2ReOb9Z4alzXCa/S734lGYRSUdIek7SeDnfl/SapCclrZfSLSzpu0W6FLaypMuSO/zXJZ0haf4ull29shot6VRJj0qallzAPybpNEmLVaQfKulLkv4p6egUtp6kP0uaLOlNSZdKmmO4WXdkO3gtw1L+EyVdUCfdiOQu/QFJ76Zr/oekvWukHyDpMEl3yV2oz5C7Wv9dMuLytF8CbsfngwJ8tFRHRqV080naQdI1kur6J0917TpJL6d79JSksyStWOf6viLpkaycl5J0rqRXJU2SdImkHjUyUn3/sqSHkt6PSzpK0qAa6Yem56O4L9MkPSP3nlt3Xq2kzST9QdIrkt5P9+dcSUt3Qt/RqvbQO6GTl17Od0fgPrIhmKX81yilHyLpEEl3pud/SnpmjpE0sju61NEx3gMNfA+k9JK0paSrimdcUj9JX5X0b0nvJR03L8kdA/wJGJyCNs3qyttdKZMgCIKWwMximws24AJ8OKYBi9VIc0OW5sc10gwALgT+BmwIjADWASYkudeB1VNaAf1xV/NFvsulsP74vIZF8GGaRfy47Fz9gMVwRwizxeON+t/hc0GKuPHAT7LfBvwf7rVtcind7sBU4EVgWhZ3XoPLfUw6x3Rgj1ReqwP/SOd7Ghia0g7Eh5G9nelzNLAz8EHa8mt7KyvrLstW6DyrnErhpwCvZvEX1JBfC3gJuBJYGXeJvg0+z9OAcyrq1M0p7njclf9HgF+lsBnAGqV60R84NsU/m9Wp/inNPsAjma7P1tB1KHB5Kre98fl6HwFOwIcTTgf2Lel6MvByqZw3xuv+f4ApWdzNPVCfirw3By4p3ddiux0YVpJdAPcM+iHuGXcUsCJwTZKZBCxRcc75gNOzMhqJfwT6fZJ7E1i1JDMu02VMKW7jFD4Vn9+0IjB/N8ulqBP7ZeedrU5kaZdL5fBQ0nMY7iG4KIf/AGvGe6DPvwf2AG7N9HgWnx9+Y6pb+TM6HVihVKf7A79J8bdm9WW+9u7nMEZaEARBowEmWgP/e6q2Hs08tt7baMewAw7P4p8CRtbI57TUKBpcCh8CvJDkHyVrTFGnkZfiV8nix1XEf6Icnxowo4E9s7iLcCNuCdzomw58GZgf7zUs0l0OXF380eNfbYsGwkxg0QaW+xXFOUvhy2f6fC4LH4GvtzYzxf0W94S4EW4oL01bA9Rw46V/d2VLutVq0A1KDadJ1GjQ4Q3YN4G/A/1Kcb/J8t4yC/9KCnu9lL4f3nM8RyMwxR9NDaMt6Sq8kVfPsCsMo23r5D8T2CELH4Z/cCjK+U/4R5G1Uvx8zG5wjW1gfRqT5fsPvNE8LtWn8czemP1VSfaUFH5PKXw4bR83jqg4Z/GxZJtS+DLZuW4vxY3L4sZk4f2As4EngZUbVS5Z/uOL89aIH4Kv0TaZ0nsw3be/0masLhXvgb77HiB9uMCHZFuq+1fja1cOTHHbZ+c6vULPC1LchM7czzDsgiDoCegFwy6GYs7FpOFIm0v6A/DTFHwLsKn5ekbl9CviX/rPMbNpeZyZvQfckX6uBGzWCVXqeuzEv77Ohrn79zfwBkrBKsDhZvaymX0ONz7PNl/u4aFSfrsRwphiAAAgAElEQVSa2RMpr2m0OVzoh39pbhQfT/tJJf2fxL+qgze0ivB3zOxf+BdxgLHAdmZ2e3ruX8DXHnwgxX8M733slmxHMLMZZjYFN/xr8XN8jtMxZlZ2oHF7dpwPPatVRh8C96afHR7ul+lqtF3rHEjaAfgs8LCZ/bkiyfHAMyRjpBjeaGZTzewVvOEK/pFhBzO7L8XPBE7K8lmnM7p3ghvM7NNmNsHMnjSzC4BPZnp9QdJHsvS1ynkK/jEGSuUsaS3c8+gdZnZ9Se4ZvEcGZr+flaTyuwRYE9jAzB5pT6YHOBY3pi5M93AW6b4dhDf0F6SxTljiPdBGQ94DZlb8LxTv9hHAUWZ2rqWlTczsWtqWbuip5zAIgqBlCMNu7uRBSZPwIWN/BT4NXAxsZGabm9mLNeT2xuvEMWmezWwbsGOWdtWevICM3C33GZZ5c0sN+6p0t5biYPZGSiPd55+CG7znV8RNSfuq+VDFmmuXmdnreYT5emZHZ0HbN1C2I7xXFZjmWu2ID/eaUJHkXLyH9QvM7rTgPLyMTq2QqVdGXdY1cUja/60qMjUOL0g/l8Cfk5yiTv3D5lwjr6fqU86EcoCZPUvb/Z0PH/pWcDpwF3BGRV61yvngtL+xhg6bAocCu9RTVNJC+LtmKN7r/nq99D2BfP7sF9PPWvf8cdrKdVdJS1al6wLxHmij0e+B4jl8w8weqogvnsVuPYfZnMKJH9CyS2IGQTCP07/ZCgQ9wo7Ac8AB+Bds8LkMd7Ujt0HaH4QPXazHlHbiG4KZfSip+PnfOulmZumqeCc77qoRUXXe8/AGCzBr6Yd9gM8Di6fgrnxAuQ5vuA3E56/0lizUdmW/Sdq/XmHoFIb2ZRXhE/GeJsAdKAC74g2/jVJwVz8yVeoqX9h8q/Tz1ao0iQnZ8aZ4j1NBTZfwZvZOVt8aVp86yGX4kGTw3phCp+vwew94jz3eYzmett6McjlvmvaVH3vM7KnsXLVYDvgl3pO4szXPlf5mtDXu27vnm+FlsTEN8JwY74E2euA90F59Kt7t3XoOzewc4ByA4RpV/jAYBEHQEkSP3dzJm2kY0nFAMQRtY6q/lOYUw2ZkZq+0s7U3vHKeQtIYSWfhxvMQYAfahrF1mtRgeiL97JQXv+7ItkPRuzGgK8KShks6Avg3Xh+/CfyhQbqVWYa2RYnrrbn2aHa8eM1UfQgze4224WzDyvFyT7I/xoeoLY8bGLU+6nTrniaOAlbA57lu2418ukvuAbIp9zzeA+3Ty++BIAiCeYow7OZi0pfTvXGnJwCHStqrjkjxR716jyo2F5Fcb38Lbyy+jXsPPMHM6vUYdJTJpX1vydaieF+MljSiM4LyNaUex7/Mb2JmXzWzhxuoW5mh2fHoOuly1+eTaqbqexTD5N7KAyXtizfmFwXWNrMj0vDNWhTdjst2Q5dDgYfx+vE7Sat0I6/u0LR7Hu+BjtGE90AQBME8RRh2czlm9iY+3+GDFHSOpFqG2xtpv2u9PCUNlrRmZ9ToRNpW46e4I41fmNl3zGxGewKdoBhaVHeNth6QrUXRSBW+xEVNJG2SHe8GXIu7K9/JzP7TQJ1q8UJ2/Ik66fLxu0/UTNX3KBrUsxyUSPoaPmfwBjMbbxUOkip4Le23qpco9UTVcnAzBXe3/xbuTfGaNOeut2nmPY/3QAV94D0QBEEwTxGG3TyAmd0BfCf9HAJcmeaAlLk77Vdqp2dvX9zddUfJ52EMbCdty9TJ1DNxaPr52y5mU+96l0n7m3pAtivkw/kOr5VI0kokZxuS+uHOPAT8zsxqzpNsJGb2EvBY+rlhnZ6FhbPjlhgOlgys4fjHmj+nsAXxtfmgc3WxuKerSarn6fZofHmRStJcvD3x+VDLApdL6u053BNoG4JZb0hocc//Q5un3y4T74Fq+sJ7IAiCYF6jZRrRQbvMlx3P4UXEzH6CrwEE3vD6fUXD6+Ls+ExJ65XzkfRRfEmE3H18/ic9LEs7OB0Wi+mCL1ZcZrvseEhFfEHN+poaDh2lrpeVTpD3Ciw62wncs0ZRvvm9KbNgVaCkdVLcW8zu0KNbsprdw0ytclBVvJn9G7gv/dxE0mEV5x6GN24vSkEL01Y2i5bT0zb8t6qMinqV1ykVyxLU0zVROP0YiK99VcXaaX+zmT1aiutonWpUfeoohev6M9N8O/A5bkW5dKacL8qOz6nqaZO0J7BAyfNi1TvmJuAb6edmVHvn7C6z3jWprhXHg5Mxf2UKWl/S2mXhRBH+ywpX/V0h3gNznruR74HuPIdV75D5kuOWIAiCuY4w7OYe8vWCFq6R5gvA0+l4K+CSvJGcvJZdmH6OAG6TdLKkjSStK+nr+EK4J5mvHVeQz/PZBkDS1sCBKd9p+LwKgK9LGpsa6B+TdBFtXtYANpU0v6ThKZ/BWdxSda4/b5AOrpmq7doawfPZ8YmSFgOQtBG+XmBxH5aSNEzSMRV57CwpnxtUNLqKtF+r46imK7K5E4XhNfItPAtWldPXaesVOU3SeZI2k7SOpANxhwiPWVrzDXidtp6eQyStm/RcVtJ5tLnRXyLNU/pB1mAv6tVCRZ1J15YvtVFP17OAO9Px9zT7mm8FB+OLdx+YB6ZzFQ3m3qpPZZYpB0haCjgSfw6PzKLyunikfE1KJK0m6Y/42nLgdXGApBPT7/+jrTdneWCipP0lrSlpW0m/wYd3frekSu5aflY9MrOf0fYO+ZKkozp2qR2m6l3zRdrc+X89S3NGuQEvaRTwObye/rhBOsV7oGffA8W7vSvPYVEXVpG0VPqYeSawSDt5BUEQtCbWwyugx9ZzG/6FchRusL2P94oZ3lhbFuhfIbMW/gdbpH0EX97gY7ihPwR3k201tpMr8uyHu0s3/M/+BXxR2WFZmv1L+fw37U/Eh3XmcTcCa+AN6+9l4S8AGwLzl84/Avhhlu4ufKHb/qmMRgI/yuLvxZd/6NfN8p8P/3Jd5DsT7518Ge+xuCeLewP4ZCb7bAqfDPwDdyYwHO95uTjFHVvjvF2Sxb9aH53pdH8q+/4pfhDee1rcm3dwd/iDSvl8MUtT3v6CLxyfp/9FKc1beB08FDg5C5+KN0ILuY+nMjV8LatXgN9mZb9SqhOF/CFkdS6lG53do2dxL4UjcBf9v8UbnBuVZIYAX8ryfSqda0CKH0rbQteGLy2yMhXPWxfq1DDgetqekVPxZ3MosDX+YeYvwIIVstdkOn2YynkysAdweRY3Gdi9VEYTa9zPd4EdS3X+I5mOlu7vKNybLknPPI8zcE+K3XreUt4LJp2K8nkJN576ZWnWS/fVgJvxBbyHpvC78ed/yQa+h+M90APvgVSuqzD7M74vMDTlNwDvLX0xi98fGJKd81NZ3Lup/CvLM9+GMdKCIAgaDTDRGvTfU2vr0cxj6+Gb527Gq/5UZ2015A6qkX7LFN8P2C81Fqam7W/AbnV0WQ94AHek8AcqGk7pvE/hHv3uJjUuU6NiKr4W1sdT2FJ1rmtCluf8ddKdAoyrE/+VBtyDJdP1TsIbk+cAo1Pc5/FG1z3AeiW5Z5MOx+Luvh/FjZc3gT8Bm9U5Z5dkqd0IOyXF314j/s6KvNbGh729ifd43Yv3es3ReE/36GS8ET4V/3CweoorGmbPAftWyO6X5F5PeQxM4cfVua9lw38AcBhu8E/GG6r/BH4ALFJxzldq5HsFc36EmKMcG/Rsb4gvdv1Mdm+vx4dhqobMKODXuEOUt/H12cakuC3wRu0jwHYVsoPwebgP443tV3HDd8VSuu/Uuf5dcCOxVvzhDSqbHXBnIJPS9Y6sSLMw3iP3CP6+eQ1/h+1frh8N0ineAw1+D+Af/WrVpWH4nMqquDdK5z0Cr/svAv9Ljecn38KwC4KgJ6AXDLviC2sQBL2IpGeBjwLHmNnRvSUbBEHfId4DfZOxY8faxIkTm61GEARzGZLuNbOxPXmOmGMXBEEQBEEQBEHQ4oRhFwRBEARBEARB0OKEYRcEzaHwYjesbqrGywZB0HeI90AQBEHQMMKwC4JeRNJwSfvjTiYAdpO0emlttobLBkHQd4j3QBAEQdATlBeoDoJ5hrS+14ZdFP+7me3cyfONxT3j5YzBPTTeBazfE7JB7yDpF8D/dFH8eTNbq5H69DUkvdEN8ZPM7KSGKZMR74EgCIJgbiG8YgbzLJIWBAZ2Ufx9M3ur/WTBvIKkkbS/iHItZprZ643Up69RLNzdRaaa2dSGKZMR74GgTHjFDIKgJ+gNr5jRYxfMs0SDLGgkZjYZX68sqMDMXmm2DlXEeyAIgiCYW4g5dkEQBEEQBEEQBC1OGHZBEARBEARBEAQtThh2QRAEQRAEQRAELU4YdkEQBEEQBEEQBC1OGHZBEARBEARBEAQtThh2QRAEQRAEQRAELU4YdkEQBEEQBEEQBC1OGHZBEARBEARBEAQtThh2QRAEQRAEQRAELU4YdkEQBEEQBEEQBC1OGHZBEARBEARBEAQtThh2QRAEQRAEQRAELU4YdkEQBEEQBEEQBC1OGHZBEARBEARBEAQtThh2QRAEQRAEicfufbLZKgRBEHSJMOyCIOgVJF0o6R1JBzdbl7kRSQtK+pqkxyQd3Wx9usrcch3dJZXD1+uVg6SD0zN1US+r12kkbSDpP5Luk7Rgs/UJgiCYGwnDLgh6GUlbSfqVpKckWcX2QWqsPS3p+tS4G9ZsvbuDpNHA54HhwEFNVmeuQtJASecAzwI/AVZsrkZdY265ju4iqZ+ks4EngVOpXw4H48/UXpIW6g39usFewGLAmsBmTdYlCIJgriQMuyDoZczsRjM7AFgFeC6L+iqwAbAasBtwM7A13rh7WtK6va1rozCzN4CLgKnAL5uszlyFmb0PHIY3mK0nziFpq57IN6c3rqMVMLMPgUOBT9B+OZwJTAEuNrM3e1q39minnlwE/Af4J3BL72gUBEEwb9G/2QoEwbyKmU2XNBH4aAq6wsxeScePADdKugVvEC0M3CRpbTN7ognqdhsz27vZOsytmNk04ClJbwEN7bmRNAb4DnBjI/Otoievo5UwsxnAy5LewJ/9WunOxI27piOpH67LClXxZnYnsESvKhUEQTCPET12QdBc3q0XaWYXA1eln8OBo3taoaClmdYDeX4PmK8H8q1HT1xHKzKj2Qp0gr2A5ZutRBAEwbxMGHZB0Fw6MuTsqux4y55SJJgrmNnIzCTtBOzXyDw7SEOvo4VpiXJIvbqnNVmNIAiCeZ4w7IKg7/NKdrxA07QI5ikkbQtc2mw9gr6NpOXwYbrh6TIIgqDJhGEXBH2fJbPj5+sllLSzpOskvS5pRvKsebqkmnNbJC0q6RRJj0qaJmmypL9K2r6OzHBJ35P0T0lT03aXpAPTXJty+oWSd89Hc9ftkj5fwzPo1JL8lhVpxpTSjJH0y+RtdLqkNyX9uVGOP2roMD6LX6NefEqzhKSjJb0kaVwK+5yke1PZPy/pZEkj2tFlz3SP3kxyd0japh2ZoZK+L+kBSe8muWeSh9YVSmmPAf4EDE5Bm2bX9HZF3qtJ+q2kFyS9L+lVSVdIWqfR19EdOlMGKf0ASZ+VdJuk81PYxyRdlnSeLOlaSWs3UraD1zJI0v9IukU+F7deuq+l5/OddM33SzpU0hxDbOXsK+lWSZPS/XxZ0h8lbVlKuyNwH9kQzFL9XyML31TS7yRNb+e6Nkt15/n0HD+f6lZlXZI0v6R90vVdkMIWkHSqpBclTUnlPKbeeYMgCOYKzCy22GJr0gZcgA/HNGCxGmluyNL8uEaaAcCFwN+ADYERwDrAhCT3OrB6hdzWwNvAWcAy+Dy+z+NDwAw4skJmZeAp4ARgWWAUsA/wVpK5Fuif0vZP1zg5u4ajs7z6A5sAz2Tx5wCDS+cUMBp4FHcss1Ip/tPAi8D+wKK4MfzDetfRhXslvFfiuEzX8RU6HlWOB8YAvwc+yOLGASel4zzccM+Bw2vc5ytS+u8AiwCLA8cC76dttjJOcgsA/wI+xL2vjsLd6F+T0k8ClsjSz5fuzW9S/K3pd39gvlLeXwGeBj6DOzxZFvd8WlzXPo26jm7ev86WwQ+Al7N7cgGwHu6F8r+l+zUd2LoRsiWdn61xP7+VxRkwoYb8MsC/gduAtYGhwPq4d0oDriM9q5nMRZnOi6ftmBT2IbB9lrZfqhP7ZboU9aR4B2wL3Jtfcw1d+wM/x+cVHoY/S4sC30xhMyk9x/iz9kSpnD+O18c3aHsnGfBY+VprbcMYaUEQBI0GmGgN+k+rtfVo5rHFFlv9jXYMO+DwLP4pYGSNfE7DG61lg2gI8EKSfzRv2OCG33Tg0or8bs0acitk4aNwI+yoCpmdMl2PysIHAR9JeVU21lNjs5D9do1r7Ae8BGxZCi+uY5MKmZ9k+W7coHs2NMtzfEX8sHI8MH8qh4OyuDNxo2K5lGa9dI+K+F9V5H1+ijuwIu7oTPboUtwpKfyeUvhw3FGJAUfUqZ8TapTFLvgSFitVxF2ZZGcAyzbiOrp53zpVBvjHkSVxg8+Aq4F7gB1ww3c08KtM19eABborW9Lt2Rr3cxBuHD9S6/7gz+pTaRtaijsmO/f+WfgOWXhZ5vYUfkPFucYXchVxg9L+3FppUvwJKf6gevkDB5eetcHAm0U54MvEbJ6lOSmT3b0jdSUMuyAIegJ6wbCLoZhB0MeQNETS5pL+APw0Bd8CbGpmkyvSr4j3QJxj7i5+Fmb2HnBH+rkSaWFgScIbloOA71eocXuRPf7VvOCbwNLA6RUyN2XHB2Y6zDCz5/Gv55WYu0L/e/r5uRrJtgPeNLObSuE/BR42s9s6qlN3MLP2PJlOrQibbu7C/q9Z8ILArmb2VEpzF967UeQ/XtJSReI0RHE83vNwdsWpT8ON5yo+nvaTSnpNwY1J8PvaYdIwvtOAv5jZYxVJirIfCHwxk+vOdXSHTpWBmb1jZi/hxhPAusDnzOxPZjbTzN4wX4/y2hS/MPDl7sp2hPRMfQA8XCfZsXjv6Y8r6uzt2fFi2XFRRh9UyNyT9p2qJ6neg/dCVyJpdbwXcjL+XirncQFt74eTJS2awqemd95TKW4B/Jm6ORM/kTYnVTWHBkv6kqSJkiZ+wPvtXlcQBEFfJNaxC4K+w4OSBuBf+4uPLhcDvzSzv9cWY++U/hhJR1bEj8yOV8UdHWwIrA48b2aPV8gcgw9xetfM8kbgvrix96jbhjVZUtKCZpYbc+9Rf22ynxV6SdqkwlA7BPhFHiB33LAhMF3SK8zJwOx41XoK9xJ5i/EsM5vN66GZPSPpXHwoWn/gU8CvU3Rxb/8vffmjJPu2pJeobnifjhuSZ1TETUn7QR2+Cmccvgbj6BplPzg7zsu+O9fRHbpaBsU9u8Gq15A8Eijmo26PGxKNkO0I71UFShpEmzE9x/qDZnajfA7t0vjQy4LL8Q8oEyqy7Wo9qatr4kC8J/MOM/tvjTTn4M/6EHzo5wlZXGE83l/++GVmb6V5oQvgvZiVmNk56RwM16g56mUQBEErEIZdEPQddgSeAw7Av7aDD2G8qx25DdL+IHwIZT2Kxtmmaf9iVSIzex8fhjeL1Hu0FPAqsEaFWJmyk432emGuwoeNLg0cis8LKs69HN7rsVtJprj2PwMHt5N/rQZjX+Mq3LADn8+IpMWAjVJYlSFeUFnGZnYdPp+KlN8Q4LN4z1nRi9HZERxF2f+a2RvZVcxI5+3WdXSHHioDzOwBSc/i8yhX7i3ZRK1yGosbQFD7Gb+uIuwZ3GAHZi06/incSNwiBXd1pE+9e/rptH+1TpoJ2fGmzF7n2lsW4h3csOuqURoEQdASxFDMIOg7vGlmr+DOOf6cwjYGTm1HrhhKJTN7pZ2tGF5VeNoc0An9ivOMBF7twLk61ThPX+qL3pRdJOU9NgcDF6ShpVU6DemAPm90Rp8mkg+vK3pb18zC6vV81EXSIpJ+DDyIezLch/Y/HNSiKPtBHSj7YvhjQ66jOzS4DAqKezaybqrGy9Yi96TbmWccAEkDJR2C6/ZZfH7izxqkW/lcw3DnOVDH+EvDuYs6s3hP6BIEQdDqhGEXBH2MNDxtb7z3CuBQSXvVESkabqt34jTFs79sJ2SK88yPexPsCX6FO7KYj9QDJ2kwPgT0zDo6deba+zqTK47zNcJqDierh6R98eG1iwJrm9kRZvZslzR0ulL23b6O7tADZVAwubTvLdla5P/tnXnGkbQu8BBu8O5sZvumObA9xbDseHQ7aYsymlQ3VRAEwTxKGHZB0AcxszeBPXF38ADnJAcDVRQ9UbvWy1PSYElFj0kx5GmhLKyW3Cal83TkXGPTPJ9OkebkFXN+DpA0P+5M5c7CyUiJQqfFJK3fjk4bdlafJpGX25Np/04W1um5gpK+hg+tvcHMxlc54ekCRdmvI2nJegmzsu/WdXSHHiqDguKePVk3VeNla5EPaay7jmP2fCPpk/hw7gHAFjXm3zaaN/CPOQCfaCdtMbG3ar5iEATBPE8YdkHQRzGzO/A1vsDny1wpaYGKpHen/Urt9Ozti8/lgdmHnh1eS0DSFrTNP3qSNs+Wh0tasFoK8DmCH9SJr0cx5GshYC8qnKZk3J0dH1sjDZJG0/4cvM5QOMUYWDdV/Xdsrbi8h6XwovlgFraLpPbOO2vh6XSfivlIv21HrjMUZT8fvkRBJZJWBnZPP7t8Hd2hQWVQ714W96zssbURsl1hIm3zzg5JTpnmQNJC+NqPBT/De+SvrPLu2hOkIdjFfNplJK1UlS55YS3eOX/oDd2CIAhajTDsgqC55A3XOdxMmtlP8DWwwBuAv5dUdnp0cXZ8pqT1yvlI+ii+JEIxd+8moPBkuLekOXrgkqOL04v80xDRS1L0osAVkoZWyB0MPFkxx06lfSVm9jBtBs3x+LpxN9RI+wC+fh/AVpLm8CqYlnY4ncY2BgsDd44hqZK2zX4OKcdn1DKMP5X2fyk8KZrZc7QZ44vi8zCrKMo2H962Am29QosyJ0Wjv8qIKhzOzMpP0nzJULgBX/geYP9032dXxnttzyb1wnbzOrpDd8qgoPJ+SVocWA0vq3MaLNveM1MZn5ZwuCb9XA5fy6187gHAebS9X6Ctx6yr9aSYM1cc555RlYWXryf/cHNIRf6FbgNxpztlI7ijbZm6754gCIJWJwy7IGgu+RpSC9dI8wXg6XS8FXBJPszRzCYCF6afI4DbJJ0saSNJ60r6Or4G1UlmNj3JvA98O8kIuFzSqZI2lLSBpG/ijhMuTw5dCo7DF1MGXxPvn5L2l7SmpK0knY+vi/fDiusonEOMqFMeBUWv3SLAGVVu8TO+TpvThf+V9FdJu0paQ9Kn8aFli5jZlR04b0d5IO33k7SlpH6SPiLpNHx+ZMFGkgZJqppPNkfvauqR/RruJOJrpehv0dYL8y1JPy3ylTRK0i9xr6UAW0taW9IawPNZHkfK1z1E0mqS/kibQ5OlJA0oGceFAbuKpKXSR4Uz8fKcRluPMsAZki6XtG2qD3vh9e4pM7u3AdfRHbpTBgWbSVqiIvwHuLHzw+Tgo4quyhbPzPAa+Rb1quqZOoI2ZyOHS7pa0jaS1pK0D76u3CjcC2tBocMekrYD/8Aj6RTgKyluIUnzy9d9K3rX8mVNtklyX6RtKYdc1zn0NbNrgSvSzwMlja24nkPw9egOqFgSoVhGZTD16ci7JwiCoHWxHl4BPbbYYpt9ww2pUbjB9j7eWDHg//Beuf4VMmsB07O0j+DLG3wM/0AzBHflbjW2k2vocnQdmQuAfhUy6+DGXZXM28AnS+kH40s4FGlexHspBtQpo3740M8pwIgOlOn+eK9BlU4P4YZII+/h1rgxWZyjOPdF+DIy+fnvBDZPcmOy8MnA+bib+6H48gET8cb49jXO+8XSdc4Ankl140jg2SzuWmDdJHdNFv4h3hCfDOyBr12W67R7dr5PZXHv4vOhji3pdEydOjQB91jakOvo5j3rahlMyML/jRsuI/FlOX6S4s6n+lnpkiz+zOyf6fQ83mM1IMUPwJf/eCfFf4AvBTK4lM92qT5V3Zv7gNGl9N8opZmU8j4eX4KjCJ8K/CyTWzDVj+JZeAm4BX+O++GG+v2Z/AnAqIpr/kuKfxP/8DEK9/B5Usr/0yWZQbjx+GEmty7uqbXIc6cs/m1gfWBgvboyjJEWBEHQaICJ1sC2SNXWo5nHFltsc27AUTUaWrO2GnIH1Ui/ZYrvhy/c+4/U8JoK/A3YrR19tkwNqsmpEXg7sEc7MoukhunTeKP8FeA3wLIVaV+sofel7ZzjMODMTpTr6vhQ0VeTTk+mBunwHrqPu+FG4zR8OOiXs7gZuJG3XklmTHb9u+DLO7yEG/gv4mvCLdfOecfiw+feSo3d24BtUtxTwO+B1Usyo1Ler6XG7aXAmBS3BW6wPQJsV3G+I1L8i8D/4stqlNNsin+YeBM3zh7Ce4RrNqC7ch3dvF9dKgPajLML8DXv/pmu8W3gZjIjsOKcXZJNdbfqmbkixV9UI/6ViryWx+cVvpLq5b+B7wLzV6Tth/fCPoO/C24DNk1xi6fyeRX4VoXsDknvSamcR6bw/an9rvtYKQ/hc4EnpLr0Lj5y4NTiXpXS31kj34kpvtZ5r6hXV8KwC4KgJ6AXDDv5eYIgCIKeRtIYvNEMsJmZTWiaMkGHkDQBN1x/Y2bje0s2aB5jx461iRMnNluNIAjmMiTda2ZVQ80bRsyxC4IgCIIgCIIgaHHCsAuCIAiCIAiCIGhxwrALgiDoPfLlIRrlyj/oWYp71pX71R3ZIAiCIOgUYdgFQRD0MHIWBQ7Ngg+StGRaeDnoY0gaImknYNUUtLmkTUprszVcNgiCIAi6Snmh4yAIgrkWSd+mbf2+TmNmo7so+g3g5FLYdrinyR8z+3pwQQlJvwD+p4viz5vZWp0832jaFl8vWABfE/FVZl9/ssjvp3kAACAASURBVGGyQRAEQdAdwitmEATzDJKG0Y1hcTb7Yu1BLyFpJO0vPl2LmWZWNrSCoCbhFTMIgp6gN7xiRo9dEATzDGZWrO8XtBBmNhlfZzEIgiAIghrEHLsgCIIgCIIgCIIWJwy7IAiCIAiCIAiCFicMuyAIgiAIgiAIghYnDLsgCIIgCIIgCIIWJwy7IAiCIAiCIAiCFicMuyAIgiAIgiAIghYnDLsgCIIgCIIgCIIWJwy7IAiCIAiCIAiCFicMuyAIgiAIgiAIghYnDLsgCIIgCIIgCIIWJwy7IAiCIAiCIAiCFicMuyAIgiAIgiAIghYnDLsgCIIgCIIgCIIWJwy7IAiCIAiCIAiCFicMu15G0iaSLpE0o9m61ELSKEmHSPqnpAuarc/ciKShkg6QdG+tMpa0i6S3JN0kaWAvq9gpJC0v6QlJT0pavtn69DUkbSfpOkmvSHpH0j8k7SFJzdYtaD6SVpB0iqQ3JI1rtj7zOo/d+2SzVQiCIOgS/ZutQCsiqR+wLbA7sB6wNDAIeB24H/gjcImZTZV0BvDLlOYYYJ2mKN1BJB0L7AUsm4L+2UR15kok/QgYDyyRgv5VI+n+wALAFsCqwL09rlzX2QUoDLqdgVObqEufIRluZwArAvsB/wV+hr87NsDv6/eapmDQVCR9FDgN2BGYr8nqBEEQBC1O9Nh1EkmbAQ8BfwKWAY4G1gZGAWOBc4AdgBcl3QQcCCwITDCzdYELel/rjmNm3wdWBt5oti5zMccCKwCT2kl3bkrzV2obf72GpHUkjaoRfTXwFPA0/mEjcA4FDgK+a2bPm9nLwOeBO1L8p5qmWdAXeBE38j/fUydo57kNgiAI5iLCsOsEko7EG9nLALuZ2Tgzu8zMHjez98zsP2Z2jZntDHwB2BAv4wXMbFrKZmJztO84ZvY+8Eyz9ZhbMbMZZvYeUHe8j5ldbWYLmtmW6Z40mxPxDxhzYGZPmtnyZracmcU4Jmb17H83/ZxVJmY2A+/x/wqwdxNUC/oIZjbTzGYCD/bgaWo+t0EQBMHcRRh2HUTSN4HjAAH7mdmV9dKb2VW4cQfeY1cwvWc0bDitomcr02fnWZaRtDGwebP1aDHWAhZNx1PzCDObbGZnmNljva9W0AfpkXdBPLdBEATzFmHYdQBJY/GvngB/MbNLOiJnZpcCf2F2w25mg9XrKT5stgLzAC1RFyQtAJzfbD1akKWz4/82TYugFWj4uyCe2yAIgnmPMOw6xk9pm9j+k07KnoQ7wAiClkPSwsD1wHLN1qUFGVEcmJk1U5Fg3iKe2yAIgnmTMOzaQdJawEbp5zv4HLvOcAtwV538JelLkh6U9F7a79KOThtLuiK5Tn9f0ouSLpC0Uh2ZEZJ+IOkBSe9Kmppcrndqjo+k0yVZxTauM/n0BJIGSDpM0l2SpkiaIekFSb9Lva7l9P0kbS/pWkl/TWFLSjo3le1USRMkbdFI2Q5ey3ySdpB0jaSac9aSHvtJulXSJEnTJT2S7vXgGjI7Sbo+uVb/QNKrkm6U9JlSunXxuT/rZsHPZPd8lyzt2pLOTuU+po6+a6e6+nTS9WVJf5C0ZZ1y2FW+5MMtKWx+Sd+T9FSqy7dKWqPWOXubonzIHCWVnpWjK2Q2S8/086lcnpf0W0mVXnQl9Zf06VQuN6ewjSTdL19O4QdZ2k1TGT+Zhe0t6SFJ0yT9S9KeWdx8kg6Vv4umyZex+HJDCqf6WpaTdFaqEzMkvZnq/SfryAyRL8lyp3xJkCny5VmOkTSyhsyCkr4t6TlJ41PYkpIuTM/OG+l4kUxmFfnyNK9LmizpaklL18j7G6msiry3kfS3VEdfkXSOpMW6WVY7y5fNeD2V1dPyd/ISWZoOP7cp/fD0PP1T/t6aKn+HHiifJ5qn7fbzKH9n7SvplnTvpkt6VNLxquHkJdXX+1La/Fl6sSLtSEm/kP8vziylP65DBR0EQdCqmFlsdTbgh4Cl7eYG5Dc+y68/8DvgPeA5fPij4cO2NqwhfyLuIfFT+IT4VYArkty7wFYVMmsBLwFX4h4vhwHbpPQGnFMhMyHFXVAKHwRcl+JuS/ksBKjJ92kAcHPS6/ik00eAX6WwGcAaWfqDcIcWxb2YgH/d/k8q/w+zuA+B8Y2Q7WAZ7wM8kuXxbA350cDfcC+t44ChuPv8h5LcvcDwksxxKe7PuBOg0cDBSW8DDs7SKtXRLTJdlkth/VP8Otl1FNuYGvp+B/gAf6aWxIcofxF4O8mdmdcj3KPsfaVyXiKFTQZey+LewJ0U9YV3RlE++2X69c+2fqW0P0/187B0PxYFvpnCZgJHlvL/Gu5NMS+X9fH3SH4fdsWXK5lVj/Dn9yr82X+5VE93BgbjPT3TgeeZvS5/vgfKah9gGu4pdklgJPDtdL6ZwN4VMsvh78Ci3g8DVgOuSXL/AdbM0i+AeyvOy2c8sGWqRy8mHYq4u/GPnjskmVeyOmr4s9k/5b0g7r22nPch6fiD0j15Dliq4prGZGnG1Xi/XYg/7xvivcH5s/c6sHpHn9ss35Vxb7Yn4EvcjEr35K0kd212rd1+HoHhwI2459w103WMww1Rw73qLlWS+V+8vv5P0m954GS8br5YSjsIX3LoDmANvD5tjH9cNeC4jtTLYYy0IAiCRgNMtJ5ug/T0CVp9wxs5xZ/VrxqQ3/gsv98C3wAGp7j18QaVAVdUyB6ON1oWKYX3A+5Jcm8CI7O4MSns72QNyhT3m0yXLUtxRYPhglL46qkRcVw5vybfp68UDZyKsnmekgGbGhQjgCdS3N9xQ3VvYGBqgORG/TRg+e7KdrCMB+GNsxupYdil89yBN8AWL8V9ITv3cVn4qln4KiWZi1L44xXnGpfJjanQQ8BRtdKkdF9OcT+uiNucNgPipCx8WLp/haH6QCqTTxd1DzdKi/N+s9n1sHRd4wvd6qQ5IaU5qJ48sxvci6atqNd34ob6R1M5v4V/5FgCWCTFG/5x57f4h4n5U14b4o1ywxu/V+GePIen+KVo+8jwUIPLZ+d030+siHuOto9VI7LwIfhzNxlYrCQzHz6iongPLpXC++MfefbIyvNC4DLg49kzd3EWfzxwK7Buihfw9Sx+9yzvIcB2WdzZ+PugMLRWxp/VIv6miusdk8WPq4g/DTdmB5fChwAvJLlHSUZYe89tih+Fez8+qiJup0z2qEY9j6l+3cCc/0WL0WYc35SFD8ffnz+vyOsU5jTsDkh5rFUKH4Z/2AjDLgiCpkEYds3fsj8xo6IB0oX8xmf5bVERX/S+PV8KHwVMAU6tke9JWb4HZOHFV+ytK2QOyGQ+X4qbQMnoADbDv8R+sdn3peJazqS2YXJVivtzRdzvUtwUYP2K+DOyMjqrUbK1yrgUfwq1DbuikXl8RdyKtBlK52bhecN2iZLMYSl8WkV+4zK5MTV03aFWGrzRNiXptEgN+aJRPRNYrRR3WYp7kTm/5s+HN+IN+H2z62FJt1nPeo341fGe0rfJGuSlNLfTZuAsWor7fVZmY7NwldL9JKWbDCxXcY7Ts3u3QUX8gVn8kAaVzWC8J+w9sg9RFfXhQ2Dpimfi9Br55nX/ilLciOw65pAHFs5k/w8YUIoX/mHNKH2gwD9wFHnfWi4n/P39UpZm/VL8mCxuXMU1zQS+WuOaL8tkt8rCx9V6JlP8can+jaqIG5LJlo2nLj2PeA+pATvWuI67s3OukMLWT7+vqkg/Gni6FHZWSr9zRfpDCcMuCIImQi8YdjHHrn0GZscN9WxnZlXz9Z5J+9Gl8N3wr477p/kas234l9KCVQHSXJAd8eFAEyrOdS6wJ97Dc2k9XSXtA/wB2MfMfl33wprDefhX8VMr4qak/aCKuGJ9uHvN7M6K+KOzNNs3ULYjvFcnrrjfN5YjzOxxYBPgq/jwR7K01+M9ly+XxOqVUUeop+u+eN19zMxeq5Hm7LTvhw9jyylcwT9pZrPNqTFfA+y59LPV1uo6EG8I32Fmtd4t56T9EHxoZ05RLneY2az1MdOfR867aT/JzJ6qOMejmewdFfHPZceNKuPd8F7He8xsckX8QXj5bGtmL4DP5cKH7oIPSZyDVPcnpJ+7Sloyi343O55jPVEzex3v7QR/pj8oxRveSwolh1g2+zqT55uvU5nHv40PHyzozPtgb/y5OKbGu3/HLO2qnch3X9xYfbQiz6ezdEtKyj07d/V53Dftz6txHfm8vOI63kj7nSV9V1LhxAwzewPvGc0p0p8hab1S3JW40VeJfK77REkTP6AvLBsaBEHQefo3W4EW4O3seKFeOF/RIJi/FL5B2v8QHzbXkTw2SfvXrWKB69RQuaw9hSR9HzgG2N7MrmsvfTNIDdtZzhYkDcDnGH2BNuc3nf6QYWavS/oH/vV7KUnDzGxqO2Ldlk1ULjmRGquFt7s5nAekc9+O9/bkYZPwhbHzvDYC9seHkoE39LpCveUxPp32r9ZJcwfeYBwEbFqKa88V/Dtp31WjtFl0pFwmZMeb4kM3C4oyb++DU3vl116dnJYdD6yZqnMU97hW/X2HNmO/YDPajIX2ymwz/HnfmPTRysxmSu1W76nUf88XZdGVcrgK97AMPjyzoxTv/oPw3sB6TGknHgBJS+HDbF9ldoOqFvn/YFefx+I6dmZ2w7Hm+czsSUl/w+/j8cDekn4EXGq+uPv4ktxF+BzNJYE7JF2Gj2p4MBmh36t1QjM7h/QhZbhG1TQAgyAI+jLRY9c++QLCi9ZM1TiKP5RyC6TwptbfzF5pZyv+WIuv1QO6oc+KwJHp+AhJjWrY9QjJw9sRwL/xxsA38Z7G7vBwdlzpca+HZGuR90J0+t4mr3R7Sbofn+N5Me6goKdYPu1rGn+pd6Ro7C3eg7r0CSQNw+e/Qf1yeZ62DzVzU7l05d20fHZc70PCo9lxnykzM3uOtl7DzrwLine/OvDuf7duTnPmORJ4tQP5NmJd0+KcH3bgfNMzud1pG5mwMm68PSHps+UTmNmj+PzAV/H/0D2BB+ReVldowDUEQRD0acKwa59bsuP11IFPvj1E0QBavRMyxf0dLWlE3ZS1eRwfEgXuaOGXXcynx5G0La7vRsAmZvZVM3u4HbGOMLnGcU/L1iJ/bpftjKCkFXFnGt8HvmRmu5pZ4aSlpxia9uXhxWWKXoFJPahLX2FYdtxeuRT1Zm4ql6IOd6b+Ds2O65VZ3rvU18pscmnfEbry7u9onvPjH+96gy5dh5m9ZmZb4wZe8T5fBrhEvjxF/1L664GV8B6+ojd6R+BfknbuqvJBEAStQBh27XM5bcNblgDWbpIexdyBbVRjfbICta3/VAxXEu76up7MJrXizOx83CU7wH7/z96Zh9tRlPn/880OISFssmtYRYfVREBFBURkU8GFUXAgg4AIKILCj3EZwEFAEDcEB1SICOoIoqigImJAR1ACuA2bIAjIIgESkrCT9/dHVed0Ot3nbufcm3vz/TxPPd23qt7auvrceruq3pL0kZ6LO7hIeifJNPe9wNsi4qEOJl8sKXqkj0spByrbRHkZ2pvbRVQ622xUvt8E+C1p0LNTRNzYofL0xP35umkPM77FR5O/drk8ywJzaC3r27yHuCOxXYo+vJXSYdq15HPTit+z+0tB7dqs/PFtWWuz4veg8WzKGorf/n3aRZK0gqRt+phmb9KdLqkTy5x7W4+1JW1Q9Y+I75OOtXgfyYgN+f7ImrjzIuITJKM0Z5FmeMcD367suzTGmBGFFbseiIj5wGdLXv/RF3lJ4yVV94r0h9/n6xSSRcSm/HYDik3j5YPRG5UxpYPN2x6KTlqyVxh7+Zykt/QQf9DIisvZpAHdt9sYomhHu3ehmFW4uguyfSYi7iGdGQVwQMWwwWKyEvXR0jKq4ny/q2uMp3STot+MJ+19aqIY4A906ewyT+6jheGHDfI7uBTZWETxfEdSuxS/TWOoGZiXmEFrOeUsWkswd6+LnCn60UOkvZuDTe3vgdLB6cWz7MvvQfHb/3JJ+7eJdyBJkekNd9EyFPORpt+QzKdJBrgGSlGPXfPe3iaOIu+llLS1pMXLxCNiUURcTDKuclv2XrwkU9KZktYuxX8sIj5MWp65iGSEqGxsxhhjRhRW7HrH6aSZDoB31K3tryMrHOeRrE8u9u5tppVln5fQsrB4Qt2SEkmr5LIWxgJuJR0eC/AGSUfVyKxEOtuqapBliXLmgei+pMNsRwPfldQXAwDdZA1a+x/r9kEWS4BG14QVNClHZWMeX+mwrCrXvoRfnK9TgPOKWblS3iIZaihbDyxmOdq1UaFMlCkryiuV4pVnjtVwD+koimJAXrV4WaS1Kq0zF/+nEtzb36mhWibdxOLZyepyscxZpfvadiE9s3GkJcb9+bAAnW2XTqV1CenMToDjaiwYImkz4FjSTDwR8Q+SZUOA7SU1rZ4o/L9a3hvW4WX07dJqUpJ2zXK30/rYUZdeNe2LS/fnNLTVy0hWcH9a8m58b7PhrO9krzWBSyWVl7oWcQ8nWb8s77Hr7/t4ccn/u5I2roSTn+nOEXFLyXu/6rOLiMdIxzXAkkt0R5MOMqcS/wpabb5UPY0xZqRgxa4XZMMOewA3Za+LJH24nYykSaRDcC+rLHmbXIrT0z+YxXHzDEsxczgWuEzSeZJ2ljRN0iEkJe77lWWIx9AaVH9R0vmSdpL0akmHkYyM3BERN7MkhfW5SaUyPE6a2Xs6h/90GVnW8iitQeIRkrYFkLShpPNpzUaukw2HnJAV2jKvbFjGdDRpoHZBw5EGA5EtDChMqgmD1jOo2x95KukcMEim42dJ2lvSNpL2JVnE3IWkVBUUptpfl/sLkqZkYzOnlOKtJ2kvSTvmvx8vhe2W5XaltfeyXNalyhsRfyKdpQbw1oZ9LoXp/49k651lCiuFbZcgV/NdBlirdL9eNTAPNi/Nfx4maXpNGkfQOpuyOhNdtMdSaVco2qWpny22wNvwm1RehteURp/IRwsUfW4CcI2kT0naVtIbJJ1E+r09vWJI4xha/fFsJeu3i5E0BdiP9LtWXmkBS/aPntqi6be5aIt27fCvVUU+z55/kvRb/MGaIynKxlSWSDtb/P1W/nMycJ2kM/Iy620lHQPcyNJt1dN7ezKtmf+dgD9IOjj/hrxZ0gWkvbj/VSlrf9/Hy2hZeV0XuCk/8+0kvU7SCSTl6/iK3JbAx2vSL5599YPHifmjQF38YGml2hhjRg7R5YPyRpIj/bP/HGlZSpCW+RxEsta2ImlD/6tIRwNcQ+kQWtKgdSPSYKU4hPVT5MNhc/hU4JZS+H8CkyppfKMUXnXfBUbVlPv9pK+3dTI/B1YoxV2JtAeiiP8YsA0wLoePpXWwdnGu02506ODiATybsyr1epyk7H2YdH5U4b8AOLokNzP7zyft4Xk3aeCyJmmA8QLp7Lel6tdfWdIMzB6lfjSPZPBlfOk5vzynWZT7CGClSjrTSUpt3XO9B9i4Ev+dlTjzchnOz2Uv/BeSBmGjstwokln6IA1M7wf+kvuKSDOmPy3JX5j9VMp7VKm9niYtv1sju2PzszqyUt4xpAOKF2S5Z3JfWyGHjycdcTE/hz9POh9s/BD3xbGk5dBFmwXpMPu1qBxEThoc/5zWu7Y/SUlelzT7vhB4R0VmAsmQ0ZxS+h8FVq1pv1eS3tEi3kHAxFL4hqTVCEX4f5A+Rig/s3Vpnf8VpN+f1TrUTqNo/3tWe5h0btui319Deg8mZv/fk35j163ITCItKSzSvhF4afE8SErIjFL47eTZ0tJz2p10JEeQjLLsUITnOOX36iek/wUTSYrJVaTfg0Nq6jOFtKqjkP8ZafnpqFKcFYEr27TVGQ3tW/veluK8mqTc1aU5F3htJ99H0vt+Y0N+i1j6N2DrUthM0v+iyaQ94w8AfwZWKcX/Iq136XDgZaT/yx+nTZ+qOh9QbozpBgzCAeVDNvgZzo40GDqetEfm7/kf3NOkZYrfI5lYrg7gjmzzT3lz0lLI2vCa/PcmfXWcRzKFPhs4lBqlriQzjTRAeyyX9SbSl9tRlXjPNJTjDzn8K02DgCF+JhNICtw/8sDjSmCrHLZ5HgT8HTiwIjczl38WsBdJWX8qD06uJ53xVtuu/ZUlfWGua8PZOfzkNn1lQiWttUj7C+8nDTrvJs3mTWko80GkQevTud+8M/uvQLKW+URux7EVue2AP+a6fZ88cCbNCjaVdbea/N+en80/cxnuJC1X3rwm7ncb0p2Tw+9t145D2BcfbtMmAaxXiS/S/qhZpPdzIcn635nA1D70nyjHB05rE28KrUFwnduetC+3KXzjDrZXMcO8gHQG2lXALj3IrEGakbuN9M79k7Ts+GCWfkcmtKnHF3NbNIX/LKcxtyH8olI+hd8HSLOR95CWzz9CWl68TU091muTd1XJGUV6f3+b22pBrvM727RT7XtbifMS0oz630i/IQ8D3wQ27Mb7SPqwdQzpI+ZTuW1/TjLoVC3b1jXpvUDaI/hf5I8Upfh1ffop0v/qt/e2T1qxM8Z0g+rvYTecUj7GLJ9ImkkaVF8bETsOlqwxZmQhqfhn+u8RMXMoy2IGxvTp02P27NlDXQxjzAhD0k0RUbftomN4j50xxhhjjDHGDHOs2BljjDHGGGPMMMeKnVneKazfVa1kdlvWGDNCqFgT9e+BMcaYIcGKnVkukTQhH5K7c/baIh8X0OMZRwORNcaMHJRYBfhYyXuGpI2qRzEYY4wx3abu0Fxj+oyk44Dj+im+KkueudQnImL1fojNYcmzqsYBP4DFB/g+Uys1cFnTZSS9h+bD5HtiCslKZfUsvd6yTUTc30/ZYYOkOQMQPz0iTu9YYYaWd5IOWy8zjWS18X+A9wx6iYwxxiy3WLEzneIc0vll/WEMyYT1oBER/V4uNRBZMyhcTusg5L5SnN/2Yj/lH+2n3HBj8wHILuhYKYaYiLiU1GeMMcaYIceKnekIEVGcq2TMkBIRT5POyDNdIiIeHuoyGGOMMWZJvMfOGGOMMcYYY4Y5VuyMMcYYY4wxZphjxc4YY4wxxhhjhjlW7IwxxhhjjDFmmGPFzhhjjDHGGGOGOVbsjDHGGGOMMWaYY8XOGGOMMcYYY4Y5VuyMMcYYY4wxZphjxc4YY4wxxhhjhjlW7IwxxhhjjDFmmGPFzhhjjDHGGGOGOVbsjDHGGGOMMWaYY8XOGGOMMcYYY4Y5VuyMMcYYY4wxZphjxc4YY4wxxhhjhjlW7IwxZpgjaVVJx0i6Q9KJDXEOl/SkpIsGuXh9RtJrJD0k6WZJqw51eUwLSeMk/ULSE5L2Hury9ER+N45u924YY8xIwYqdMcMASe+TFD24nwx1Oc3gImmUpHOBu4AzgU3bRD8cmATsL2m1wSjfANgfWAvYBthpiMtilmRLYBdgCvD+IS5LI1kBPQ+4F/g87d8NY4wZEVixM2Z4cDGwErAH8HjJfy5wILAO8I4hKJcZQiJiEfBhYHMgeoh+DjAfuDgiHut22XpC0pvbBF8EPAT8AfjV4JTI9JI/AVeTfnu+McRlaSQingOOIn0c6OndMMaYEcGYoS6AMaZnIiKAhcBPJf0QOCgHfTkiLhy6kpmhJiKeBR6UNAdYo028c0jK3ZAjaRSpLJvUhUfEDaSPFWYZIytM7ZTyZYaIeBq4W9LjwLI+S22MMQPGM3bGDD8eKt3fP2SlMMsazw51AfrA/sDGQ10Is9zw9FAXwBhjBgMrdsYMP15ouDfLNy8OdQF6g6SpwBeHuBhm+WJYvBvGGDNQrNgZY4wZFCRtBPwCsKVLY4wxpsNYsTNmOSWbAT9e0i3ZDP4Tkn4n6aOSJrSRmyrpXEl3S3pW0jxJf5Z0sqRJNfEnSzpS0m2FuXFJ60n6uqRHcr7fWdbM2kuaKOk/Jf1R0kJJT0u6R9LXJC21N0zSWEnvkXSdpAuy32aSvifpsdxOV0ia1knZXtZlvKT3SvqVpEZjJDne0bkfPJnrfIukD0saXRNfkg6UdG1+js9JelDS5ZJ2qcR9K3AzpSWYFauuW5f83yjp25Ke6aFeO0m6VNJ9kp7J1wslvboh/gRJB+T6zcx+q0g6U9IDkubndp7aLt++kvM9MOdbPN+9Jd2Zn+8hlfijcvxfSXo81+12SadImtImn40k/bekv+V38zFJP5L02jYya0g6Pb+fT+W+NkvSvzbEX0/SCZLulzSj5P9J1Vvr/UtF/uBqnJo8tszP8f7cpx7Jz7n2uZbk/lXSL3O9n5Z0vaTd2skYY8yIIiLs7OyGkQNOJFl5C2BGP9OYBtwHXAe8mmRxc3vgtzndO4CXNcjNBR4jGVCYDLwOuD3LXQ+MynHHAmcAD5bKeyLweuBR0l7B+aWwa4a6bUv1XAX4M7AI+BDJtPumwI9yWZ8A1inFP6FSz5nAdrl+L5T8A3gG2LUTspUy31u0ccX/2FJYALMa5DcAbs19YhowMfeJh7LclcCYisxFpTKvnd1J2W8RsGcp7iiSwa6DSmUZU7gcZ3fgpnKdG8o6BvgyaV/hUcDqwJrAx7Lfi8AnKjKfBP5aaedXAn8D5pCszRZhd1TrOoC+dBrpfSnnu08uY+E3rxR/EmlW83KSRcfJwI4ka5SRy7teTT4HkPaSfRpYF1gZOC7LvAj8W43M60j7dI8F1s/t+JGcTgDnleKuCnwvt+9Svz/AeOBtuS2L8E8B4yp5jsp5PQ5cC6xfCT8y1/HdJIMnGwJfzek9DxxQU4+xwKU5/HjgJbkvfhp4Lrul3o0mN23atDDGmE4DzI5uj1+6nYGdnV1nHQNU7PKA5zHgHmCFSthE4P9y2ncBkyrhs3PYGRX/XUplem3JfyXSeWTFIPYnwFXAq3L4aOA7JdnpQ92+uVyfy+W5seI/qTTo/XjJf3IeTD+Rw34I3Ajsleu4OvC1Uj3/CawyUNlK2e6tG7zmAfdY4DYaFDuS4np3dhMrYSeVC+onYAAAIABJREFU8j645L9Xyb8q85vsf1VNXjMKuZqw8fn69aY4OfzUHP7BdukDh1f64gq0lKxZwDXAzqU4p5dk39WhvrQ2yfpn8RHjSuAKkiL6SWABMLMU/wf5HRlVSWct4KmcxtWVsLeTFOnTavL/e5ZZCEwu+U/Nfe59NTIfLrXD+7KfgHHADqWwpX5/gPeUwvdtaJOVc102rfjvndvj5TUyl+U0nwU2rIRdkMMOq5E7sVSeE+vKU3VW7Iwx3YBBUOy8FNOY5Y+zSF/fz45kDnwxEbGQ9MUeYCPSbFKZV+brExX/G0v365fSWxARD5MG05AGhntFxM05/EXSYLqg7VKrQaS2nhExnzQ7CUvW88mI+AdJeQLYFtgvIn4SES9GxJyIOIQ0oId0LMEHBirbGyLi2Yh4nqSwN/Fp0szIZ3MfKPOb0v1apfuijZ6vkSn6w/r0gUhHN0A6v64WSVuRZpjmkRTeahozgf/Nf54hac3svyD397tz2CrAPhFxTUn8NFpnnnWkL0bEQxHxV9JsKMAbSArIIxFxMunjyYxct11Iys1Zkc4oLKfzMFAsa3yT8nJgSSsA55Jmc0+tKULx/FYgKVQFJ5NmhC+ukbm6dP/BnH9EOuqg8dlkLiEpkwD7NcT5N5JyemfhkZf6fhH4eUTc0aZM4ygdjJ6XWs4gzbKeWyP3RZLS2xZJh0qaLWn2o48+2lN0Y4xZJrFiZ8xyhKR1SANHgF/XxYmIX5CWQgEcKmlcKfgzWe6Sitj80v34mmSLAftv8+CwzN2l+8b9Q4PMV4DfAWfXhBV1ratnUber8mC+yidK93t2ULY3PFXnKWk8rYHyL6rhuT/sCRwGfKEUdAlpKd0pNcm2a6N+lzVzGGkm8/qIaLIKe16+rkjrzMeCoi/eEhHzygER8ThpqTF0vi8W+f4wIhYfU5K/4hYcmK/nS3q46oCtS3G3yNd3kmb/bqzWJ/NBUpvtXuSblcF3kxS9h2ryubYmn4J2z6b4WFO8N29t2K94OOkDU5kdgZcBb2moe1lpLZepeC9+XGnLojxzgX+0K3OOd15ETI+I6Wus0XgcpDHGLNP4gHJjli/eThoUAzzSJt61pBmcScCrgBsAIuIzJOUOAElrkwbO+5Zk6z4YNZobj4gnJRV/9lcR6CgRcSVpyRwAklYkLTGbQWsmp88fxiLij5LuJS2De8VgyWaaZi2mkxQggAca8r6yxu8e0mAcWHzo+FtISuKbsnd/Px62m2F5R76267+zSvdvZEmloCfT90+SZvM63ReLOrU7ouQ1+fp2Wh9XmigU0Dfma9Oze5KlZ7KmkWa+ZgNv7SGfJZSliFhUel+b+DppCeSKpD1zHysCJO1M6hdXV2SKun+D+pnHMs/mtNYiLQ0FuLM5es8zdsYYMxLwjJ0xyxflQ6HbDXZuL92vXQ2UtIWk75D2As0lLS8bcUh6iaTPkgxXbEwyUPG7ASZbLIlcuW2szss2sW7pfmxfhSWNk3QEqWzvIe1P/FKHylbNayWSYQxo038j4j5aM0tL9d9lmGKp66KIeLgHV1gMLZ5fX55dkc+kXuTTToGuJSKeAC7Mfx6UP4wUHEFaBl6dXSvKNL4XZSqWSG9Tkm87k2iMMcsDVuyMWb6YWLpfvU28uaX7xfvM8iD+c6Qv/b8HtoqIsxuWgA1rJB1IsqK4JjAtIj4eEfd2IOl5letgyTZR/j+wYV8EJW1L2vd1APD2iDgwIm7oYNmqrFS6b9d/odVG1f2gyzKFcrZVH2SK59eXZ1fks7GkiW1j9p8v5+sqpD11SFoP2An4Zpsy9aXu5SNSlpVl3MYYM2RYsTNmOSEbkbi/5LV5u+il+7uy/Cjgf4CPAsdExBeqBh5GCpKOJpmlvyoiZnRYcS2W+N01yLJNlGdk3twuoqQ3lO5fS1qyOxZ4U9kQRheZQ7JKCu37L7T6cN1+xWWVOfm6T7tIktaWtEH+s3h+W0lq3BwmaXTpPLsin9H0sBSz3Rl47YiI22jt2fxQvn4A+HZeHlqlKNOrJa1bE14u0+vybTmd6l5AY4xZ7rBiZ8xyQB7wHQP8suS9exuRYoD4u4go9u7sTsvwyoVLi4wMlA5KL/b49Lee7X5bi5mV6h6jTsj2h9m09p0dIal2SZ+k1YCDS15fAiYAl0XEgg6Wp5FsLOW6/OcGkl5eFy9bWCxmc74/GGXrEL/P110l7dAm3lG0ZqiKpcFjSPvZmphBa1nqbFpLWT9ZMZC0mOx/fA9lbkexJPdfsvXKg0mGieoo6j6atD+vFkmvAN6V//xTKWjvpnqUGN1DuDHGDGus2BmzfPAJ4Iq8TG529tsnW8msY1q+lq1ClmdI1ixHrigDdYOn3v7W9GiVYRDYhNbM2Jo14UVd2w0SV63zzMZmtiQZ0DivLs4AZFW59io8H+Hwo/znRix5/ESR91jgfNIZewVFf+hrGy02HpL3zBX3K9SUFS1tqaNsTfGImvSLso0jGdSoKsFD3Rfb5V8cPSDgu5I2rkaQNI109t4t2esS0lEHAMdJ2q5GZjPSERFXwOI9cD/Nwf8CXCCpzpjaySxpiKb6PHpqoytpzZh+E/hLRNzeEPcqoDhn4GBJh9fUYzzJEMxFuR5/p6XYrpnLW0dRzpUawo0xZkRgxc6Y4UfZWl+PAxVJbyOZNi/OszqMZFVuPHBWdeAsaUPS7NzVEfGtUtB9pfuzJE2WNErSHrTODQNYT9Kako7L6YmWslIevNcxuaf6DALlen5C0qYAkraUdDktgw3rSRor6bSaNHZqUJpPICk7/5UNfNTRX9nCoMqkhnSLGZ66Nv44LeMTH5H0Q0m7SXqVpANIZ5dNIR2eXVCUYd/cB5C0Vt6DWcwcrSZpQj4jrJhde7yUxm5Z7v0seYRDeb/UEuWNiCuAS/Ofh0maXlOfI0jWHA+pORJhtXwd7L5Y5LdemziX0VKk1gVukvQpSdtJep2kE0iz7otn0SLiUVpHTkwArsky20p6g6STgJuA00sGVwD+H61lrfsBN0jaT9LWkvbMff2twDmVMjY+myrZQEqhiL+EpY84KMd9miVnB8+WdImk3SVtI2l/0vmId0fETaV4x9KacT5W0hckTQGQNEXSV2m1+a6SpkkqHxthjDEjh+jyCeh2dnaddaTBX2R3HbAOSUkbk904kiI1Hfg88DxwTiWNtwELchrfIx02PRHYhfSF/WfAyhWZSSST6kXez5H2uNxBmkl6NPu/SNrLtwnJ3PmhJZm7gZcDY3OaE0nnbBXhfyeZ8h8zxG38o1KZFpGUkXmkYx0uKYXNA95VkptV8r+VpLisTDqo+/M57AJgVE2e/ZIlKQwHl8p0H2nGqmjjsaRDz5/M4c+Tzj5boZLOHiTlLmrczcDqlfgfrcR5Iqd9CmmpYOG/APhSSW5VYGEOe4F0xtivSB8aR5EG4beU5E8FptTU+ec5/DFgf5LCsS5pxnEh8I6KzHiS8rioJLctyQpjkebbSuFzge2BcQPsSxNzvs+V2v99wOSG+GuQFJi657AIOLJGZhTpmIA6mQBObshrz9KzqLoHgE0r8VcCTirF+SNpeXDj+0r63ZgH3ENNv6+Jf1KbeswCVqyReX/uS0W8Z3N+z5BWK9xbCrsC2LZdGaZNmxbGGNNpgNnR7fFLtzOws7MbuCPtjdmbtD+ladDTzu1ck+YGpK/xd+cB0IOk5VD7AqMbyvEKkkGE+XngdzowMYcdTxrEX1MMCIGHG8pzKek8tqbyfm6I23tKHij/kzTA/y4wNYe9iWTo4TZgj4rcrFz+maQ9TX/IbTs3t8u72uTZL1mSIZXaNs7hFzWEP1yT1sakfYUP58HxrcB/ABNq4o7Kz/wekkJ4HfDGUn+9jWTY49ga2b1yuZ/I7bxy9j+Y5j6xWSUNkQ70nkVS0haSjlw4s3hWlfg3NKQ7O4c35XvpAPtS0/OJNjLjSHtib8ltO5ekyO7UQ177kmbmF5AU+auAXXqQ2YB07twDJOXzPtLvzJqVeGPatNFpPeTxBeC4PrTZG4Ef5+f6DMny6nG0UbJJH7J+SPoIszD3x91y2N0kw09b9SZ/K3bGmG7AICh2SvkYY4wZKJJmkQal34yIGYMla4zpHNOnT4/Zs2f3HNEYY/qApJsiom77QMfwHjtjjDHGGGOMGeZYsTPGGGOMMcaYYY4VO2OM6RwT87U/ZtUHImuMMcaY5RwrdsYYM0AkrZiPldgie+2cTc33ZFJ/QLLGGGOMMQV1B5IaY8yQIuks4L39FJ9P81luPXFfRLyqLwKSVqd1sHLBKsC1JKuQa3VD1gwO+TzG4/orHxGrd7A4xhhjTCNW7IwxyyKfBD7TT9nxJFP9/eHFnqMsSUTMIZne7zMDkTWDxjmkIyCMMcaYZRordsaYZY6ImEc61NiYISUiFpDOhTPGGGOWabzHzhhjjDHGGGOGOVbsjDHGGGOMMWaYY8XOGGOMMcYYY4Y5VuyMMcYYY4wxZphjxc4YY4wxxhhjhjlW7IwxxhhjjDFmmGPFzhhjjDHGGGOGOVbsjDHGGGOMMWaYY8XOGGOMMcYYY4Y5VuyMMcYYY4wxZphjxc4YY4wxxhhjhjlW7IwxxhhjjDFmmGPFzhhjjDHGGGOGOVbsjDHGGGOMMWaYY8XOGGOWUSStKukYSXdIOrEhzuGSnpR00SAXr89Ieo2khyTdLGnVoS6PGfnkd+jodu+QMcaMFKzYGTOESHqfpOjB/WSoy2kGF0mjJJ0L3AWcCWzaJvrhwCRgf0mrDUb5BsD+wFrANsBOQ1wWM4KRNE7SecC9wOdp/w4ZY8yIwIqdMUPLxcBKwB7A4yX/ucCBwDrAO4agXGYIiYhFwIeBzYHoIfo5wHzg4oh4rNtl6wlJb24TfBHwEPAH4FeDUyKzPBIRzwFHkT4i9PQOGWPMiMCKnTFDSCQWRsRPgR+Wgr4cERdGxEN5gGKWMyLi2Yh4EJjTQ7xzImJyRLxvkIrWiKRRJEWzloi4ISLWiYhtIuLxpnjGdIKIeDoi7mbJj2bGGDNisWJnzLLDQ6X7+4esFGZZ49mhLkAf2B/YeKgLYUyFp4e6AMYYMxhYsTNm2eGFhnuzfPPiUBegN0iaCnxxiIthTB3D4h0yxpiBYsXOGGPMgJC0EfALwJYujTHGmCHCip0xI4xs3vt4SbdkM/hPSPqdpI9KmtBGbqqkcyXdLelZSfMk/VnSyZIm1cSfLOlISbcVZsQlrSfp65Ieyfl+Z1kzay9poqT/lPRHSQslPS3pHklfk7RJTfyxkt4j6TpJF2S/zSR9T9JjuZ2ukDStk7K9rMt4Se+V9CtJjcZIcryjcz94Mtf5FkkfljS6Jr4kHSjp2vwcn5P0oKTLJe1SiftW4GZKSzArVl23Lvm/UdK3JT3TQ712knSppPskPZOvF0p6dUP8CZIOyPWbmf1WkXSmpAckzc/tPLVdvv0hP+Ojct7z87tzf67n9Jr4oyTtmcvzy+y3bn5vHpa0QNIsSW/qpGyOu72kb0p6Ov+9vqSf5HJfLml8Jf66kk6R9Jf8rszJ6R9S7TeSPql6q75/qcQ7uBqnppxb5ud9f+57j+T+UPv8S3L/KumX+d16WtL1knZrJ2OMMSOKiLCzs1sGHHAiyXpbADP6mcY04D7gOuDVJIub2wO/zeneAbysQW4u8BjwZmAy8Drg9ix3PTAqxx0LnAE8WCrvicDrgUdJewXnl8KuGeq2LdVzFeDPwCLgQ8AUkhn0H+WyPgGsU4p/QqWeM4Htcv1eKPkH8AywaydkK2W+t2jjiv+xpbAAZjXIbwDcmvvENGBi7hMPZbkrgTEVmYtKZV47u5Oy3yJgz1LcUcAY4KBSWcYULsfZHbipXOeGso4BvkzaV3gUsDqwJvCx7Pci8ImKzCeBv1ba+ZXA30iGZx4vhd1RresA+9NY4Jqc9inAasBLga9lv2eBrUvxP0g6wmLxMwM2ys/ihdy2RdgiSr8DA5R9C0n5Lve51YA7K377VGSeAC4D/oV0pMYuwG057m+BVUrxxwNvy21epPcpYFylzUYB6+fnci2wfiX8yPzs3p3LuCHw1Zze88ABDc/h0hx+PPASUp/9NPBcdku9Q01u2rRpYYwxnQaYHd0e53Q7Azs7u945BqjY5YHMY8A9wAqVsInA/+W07wImVcJn57AzKv67lMr02pL/SqTzyF7MYT8BrgJelcNHA98pyU4f6vbN5fpcLs+NFf9JJAMLAXy85D8ZWDcPcINkufRGYK9cx9VpDeID+Gcx2B2IbKVs99YNSvNAemxpoD2rRnYKcHd2EythJ5XyPrjkv1fJvyrzm+x/VU1eMwq5mrDx+fr1pjg5/NQc/sF26QOHV/riCrnvFwrPNcDOpTinl2Tf1cH+dGRO89GK/yjSB5YAzqv0p8m0FNH/JSnc/waMy/3wv0plfRrYuAOyqwErAz8ohX8d2JF0nuDtJINN6+X4W5CU0usBVeq2JvAILeVudCX8PaU89m1ot5WBp4BNK/57AwuAl9fIXEZLWd6wEnZBDjusRu7EUnlOrCtP1VmxM8Z0AwZBsfNSTGNGDmeR9jidHRFLWIGLiIXAR/KfG5Fmk8q8Ml+fqPjfWLpfv5Tegoh4mDSYhjSw3Csibs7hL5IG0wVtl1ANIrX1jIj5pMEtLFnPJyPiHyTlCWBbYL+I+ElEvBgRcyLiEOCKHL4G8IGByvaGSMchPE9S2Jv4NGnG47O5D5T5Tel+rdJ90UbP18gU/WF9+kBEFJY9/9AUR9JWpFnIeSSFt5rGTJIyA3CGpDWz/4Lc3+/OYauQZp6uKYmfRusss072xab+tIg0QwlL96cnabXjlsBxEfGtiHguIuZHxKdoHRkxgTRbOVDZxyJiHunjS8GtETErIn4VEZsBL42IB3LY+aR3+sw8GCnX7RHSLCnAa0gziWUuAf6e7/dbqsUS/wZcHRF3Fh55aecXgZ9HxB01Mlfn6zjg/SW53UhK/x3AuTVyXyTNYBpjzIjHip0xIwBJ65C+dgP8ui5ORPyCtMQJ4FBJ40rBn8lyl1TE5pfux7M0xYD9t7H0eXt3l+6nNBR9sPkK8Dvg7Jqwoq519SzqdlVE/LUm/BOl+z07KNsbnqrzzPuligHwL6rhuT/sCRwGfKEUdAlpidwpNcm2a6N+lzVzGGkm8/qIaLIKe16+rkha+lmm6Iu3ZCVmMZHOzJub/+xkXzyfNKt1Zk1Yb/rTTRFxQ034iaU4Tf2pP7LlozO+XA4oFDhJ2wHF3sDa3xLgYtLMGqQlzeV0XqT1fr21YV/j4aQPUWV2BF4GvCXvF1zCkWZzC7Yo3Rfvz4+rSmguz1zgHw31WIykQyXNljT70Ucf7Sm6McYsk1ixM2Zk8HbSoBjSMqkmrs3XScCrCs+I+ExEvKFQPCStLekTwC0l2brfi0Yz4nl2oaC/ikBHiYgrI2L7iLgcQNKKkg6SVOxJhH78LkbEH0lLJgFeMViymabZiOkkBQjggboIuT3OLc/MRcQ9EbFjRJwIiw127C7pUlqD+P7+72g3c/KOfG3Xf2eV7t9YCevJpH3RHzvWFyNidkS8NiLOhcWGVPaV9FNgnxytP/3pUdIyR4D1JK3UIdlFpXhNynPxHIK0Z7Yuj6dozRxuKmmtSpSvk5T4UaTlqouRtHP2v7oi85p8/QawdY17Oa39ngfmtNYCdshyd9JMjzN2EXFeREyPiOlrrLFGT9GNMWaZxIqdMSOD8qHQ7QYxt5fu164GStpC0ndIS7bmAm/oTPGWLSS9RNJngT+R2u4A0kzeQCiWRK48yLJNrFu6H9tXYUnjJB1BKtt7SPsTv9ShslXzWolk8ALa9N+IuI/WrN9S/XeokDRJ0sdJRmpeT1oC+f0BJlteYtvXfjEQ2eK3pNiX1kTjb0lEPAFcmP88SNKKpeAjSMvFq2kXyuH4iHi4B1csfd2mJN9uNtgYY5YLrNgZMzKYWLpfvU28uaX7xfuC8iD+cyQjKr8HtoqIs6tL2kYCkg4kGaBYE5gWER+PiHs7kPS8ynWwZJso/75v2BdBSdsCfyEpvG+PiAMblv11ivKsUrv+C602qu4HHRIk7U6aLdoBeENEfCgi2u177C3zGu67LVv8lowi7Vdsova3pESx1HMV0p46JK1HMtbyzZr4xceHrXpd0iXPTVxWlnsbY8yQYcXOmGFONiJxf8lr83bRS/d3ZflRwP8AHwWOiYgvZOMPIw5JR5PM4V8VETM6rLgWS/zuGmTZJspLGt/cLqKkN5TuX0tasjsWeFPZwEUXmUOy4gjt+y+0+nDdfsVBRdI7ScZv7gXeFhEPdTD5ok88EhEL2sbsrGxff0uepmapb0TcRmtvZ7GE9wPAtyvLtAvm5OurJa1bE97KWHpdvi2ns0VdXGOMWZ6wYmfMMEbSGsAxwC9L3ru3ESk2j/yuZAFvd1qGVy5cWmRkoHRQemGAob/1bPebWcyKVfcOdUK2P8ymte/sCEm1yzElrQYcXPL6Esmi4mX9UAr6Rd7vdV3+cwNJL6+Lly0nFrM0A13qOCDyB5GzSQrOt9vsWWvHstif+vpb8qM2dS+W7v5Ltl55MMmAUR2/z9fRJAMwtUh6BfCu/OefSkF7VwxC1TG6h3BjjBnWWLEzZnjzCeCKvExudvbbJ1vJrGNavpatQpa/yq9ZjlxRBuoGRb39DVHPUbrOJrRmMtasCS/q2m7wt2qdp6S1SebnX6BlubFTsqpcexWej3D4Uf5zI5Y8fqLIeyzJsuMPS95Ff+hrGy0e3JcNdkhaoaasSKrWp2wl8Yia9IuyjSMtfawqLYPdF9eg1Uad7k/jaRmHaVKEBiJbKKZ1XEbLiuRBledXpvgtOachHOBKWjOr3wT+EhG3N8S9ipaxloMlHV5T5vGkIw0uAoiIv9PaG7smcHJD2sUz77URGmOMGY5YsTNm2aFsra/HAYiktwHvpnUe2WEkc+bjgbOqA2dJG5K+wF8dEd8qBd1Xuj9L0uRsCXEPWueGQbKwt6ak43J6ojW4bBr8FUzuqT6DQLmen5C0KYCkLSVdTssQw3rZuuFpNWns1KA0n0AawP9XNvBRR39lC+MXkxrSLfYW1bXxx2kZlfiIpB9K2k3SqyQdQDpXbgrp4OqCogz75j6ApLXyHszCwuFqkiZkE/HF7NrjpTR2y3LvZ0mT++V9UEuUNyKuAC7Nfx4maTpLcwTJoMchNbNEq+XrYPXFR4FninLlfYlI2lDS+bRmwdfJ79MJNdYtXylpG5bmaNK7dUGbvY39kS23zXp1ieZjSz5IaueXkI5CWYK8XHdr4BsRcV01vJRW0FLYX8LSRxyU4z4NHF/yOlvSJdki6zaS9idZ4rw7Im4qxTuW1sz0sZK+IGlKLucUSV8t1XVXSdMkbd1UDmOMGdZEl09At7Oz650jfSkvLNFdB6xDUtLGZDeONGCbDnweeB44p5LG20jnSwXwPdIByhOBXUhfzn8GrFyRmUTaI1Pk/Rxp78odpJmkR7P/i6T9N5uQzOgfWpK5m2SOfGxOcyKtwWGQDix+BTBmiNv4R6UyLSIpI/OAfUnntxVh84B3leRmlfxvJSkuK5MOn/58DrsAGFWTZ79kSYPwg0tluo80Y1W08VjSoedP5vDngXcCK1TS2YOk3EWNuxlYvRL/o5U4T+S0TwGOKvkvAL5UklsVWJjDXiDN+vyK9AFxFGlwfUtJ/lRgSk2df57DHwP2JymD65JmHBcC76jIjCcpj4tKctuSrCsWab6tFD4X2B4Y14H+dFalrR4nKXsfBs6otNXRJbmZ2X8+6Z16N0kxXZOk3LxAeldXrMmzz7Kk349NSccgFGX6NvBSavpslvlAfu5BmuHfgPRb8Q7S/s1v9qYNs8w84J6mvCrxT2roq0F6l+ra5P253kW8Z3N+z5BWNdxbCrsC2LZdGaZNmxbGGNNpgNnR7XFOtzOws7NrdiQz4XuTlkw1DWbauZ1r0tyAtDzq7jyweZC0zGlfYHRDOV5BMnQwn6TknQ5MzGHH54HpNcCm2e/hhvJcCkxtU97PDXF7TyGdk/VP0gD/u8DUHPYmkgGH24A9KnKzcvlnAjNIM13P5DSuoaQE1uTZL1mSIZXaNs7hFzWEP1yT1sakfYUP50HvrcB/ABNq4o7Kz/wekkJ4HfDGUn+9jTSwP7ZGdq9c7idyO6+c/Q+muU9sVklDpHPKZpGUtIUk8/1nFs+qEv+GhnRn5/CmfC/tQH+aQFLg/kF6R64kWZSFpIQ/QPqocWBFbiYtRWUv0iHnT5Hev+tzezUpXH2WBT7Zph1mtKnfViQF7j5Sn70PuBzYrY/t9AXguD7EfyPw4/z8nyFZaD2ONook6YPXD0nK9cLcb3fLYXeTDERt1Zv8rdgZY7oBg6DYKeVjjDGmCUmzSIPNb0bEjMGSNSMTSTNJyuu1EbHjYMma3jF9+vSYPXt2zxGNMaYPSLopIuq2GXQM77EzxhhjjDHGmGGOFTtjjDHGGGOMGeZYsTPGmJ6ZmK/9MZc+EFkzMnF/MsYY03Gs2BljTAOSVszHSmyRvXaW9IY2Z3t1RNaMTPIRETsAO2evLSTtLWliO7mByhpjjFk+GDPUBTDGLD9IOgt4bz/F59N8lltP3BcRr+qLgKTVaR2YXLAKcC3JKuRa3ZA1g0M+j/G4/spHxOr9EJtDa8YN0hEmP8jlWSEinqmVGrisMcaY5QArdsaYweST1Bx43EvGk0z194cXe46yJBExh2R6v88MRNYMGueQjoAYNCKi38snByJrjDFm+cCKnTFm0IiIeaTDio0ZUiJiAensOWOMMWZE4D12xhhjjDHGGDPMsWJnjDHGGGOMMcMcK3bGGGOMMcYYM8yxYmeMMcYYY4wxwxwrdsYYY4wxxhgzzLFiZ4wxxhhjjDHDHCt2xhhjjDHGGDPMsWJnjDHGGGOMMcMcK3bGGGOMMcYYM8yxYmeMMcYYY4wABFFxAAAeqElEQVQxwxwrdsYYY4wxxhgzzLFiZ4wxxhhjjDHDHCt2xhhjjDHGGDPMsWJnjDHGGGOMMcMcK3bGGGOMMcYYM8yxYmeMMcYsY0jaRtK5kuZLmloTPk7SLyQ9IWnvwS9h35B0Sq7LqUNdFmOMGalYsTPGmF4i6fuS3jLU5TAjl6zQXQXcDBwKrNQQdUtgF2AK8P5BKt5A+DCpLkcOdUGMMWakYsXOGGN6gaQNgLcDRw11WcyI5i/AW4D/6CHen4CrgbnAN7pdqN4g6c1tgr8MLATOGqTiGGPMcseYoS6AMcYMEz4CjAZ2k7RpRNw51AUyI4+IeB5A0p96iPcc0E6RGlQkvR7YH/hFXXhEfBz4+KAWyhhjljM8Y2eMMT0gaWXg34s/ScvKjOkmzw51AfrICUNdAGOMWd6xYmeMMT1zCPA08Gj++8Cs7BnTLV4c6gL0FkkfAt401OUwxpjlHSt2xhjTBkljgA8BZwPnZu+VGB4GK4zpKpIOAr441OUwxhhjxc4YY3riXcBLgK+SlLvnsv+Rktr+hkpaVdJHJf1V0ozst5ukX0taKOlhSedJWqvDshMkHSjpd5IuyH57S7pT0mOSDqmR2UnSpZLuk/RMvl4o6dU1ce+SFDXuY5V4v6mEz6yEj8rl/JWkx3O+t2fT+FPatW1/kLSlpIsk3S/puZznbEnHSRpXE38dSSdK+oekHbPffpJukvR0bqMzJE3upGwv67KepBNyXWa0ibeRpP+W9DdJz+bn/yNJr22Iv7qkM/NzeFrSAkl3SPpita9JOp9kuKV4Dw4sPes/lOKtJOnQ3NYz25R1RUlHSLohP5v5kv4g6aSmGXJJL5N0qqRHS+28vaSfSnoyt/+n1cO7aowxI4KIsLOzs7NrcMDvgHNLf38LiOz2bpBZFfg68FQp7gzgiHz/fMk/gL8D6w1UNsufBjxWCp8J7ENa2lf4zSvFH0OyWPgsyeLn6sCawMey34vAJyr1ewnwuVJ6TwH/AqgSbyxwWI7zMWDFUtgkkqGNy4FtgMnAjiRrjwH8rVyvDjzHPXN97ga2A1YGdgceyvl9rxR3KvA/lbbeETi94Rn8AZg0UNlKeXcsxZla6Vvfy3VZ3D8a6nwAaQnxp4F1c52PyzIvAv9WiT8VeAB4Btg3P5OtgN+WnsnEUvzRuf9cm8O/mf8eA4zOcT4HPFIq68yGsm4E/JlkFXRH0qz4lsCPstxDwDal+BuQlMpnKu38odw295M+whRh/9nbvjJt2rQwxphOA8yObo9Zup2BnZ2d3XB1wOuARcBmJb9XlQaLv2qQGwOsCOxRinsucB2wVY7zCuD6UvjVA5XNYWsDmwDzc9iVwBUkZe2TwILy4Bo4Ncf7YE09ZpTyOLwm/Gc57ElgTENbHAb8psb/B8BVwKiK/1q0lNqr69Lsx3McAzyc0zyiEnZwqY7rZL8JwHjgg6Wwc7KSsVGOsx1weyn8awOVrZRrx1L41JK/gHHADqXwGTXybyf13dNqwv6e5RYCk0v+l2b/SyrxNy7ltV9NerNoUNpyW0wCnmgTZ0Xgr8A8YK1K2Gjgl1n2MVofQCbksH1LZfsOcD6wbo6zOnBbDptLVjZ7clbsjDHdgEFQ7Lw0wRhjmjkGuDIibi88IuJm4Df5zx0lbVUViogXIuIp0jljBZsBu0XEH3Oc20gzRg/m8DdJ2n4gsjnsoYj4K3BrDnsDcFhEPBIRJ5Nmh2YA5LIfSxpQf62mHjOB/81/niFpzUqUz+XrJGCvqnzmYODzZQ9JuwB7A2dFxKJKng+TZm2Kem3SkG5fWJWk2EJSMMrcWLpfP5fhmYh4lqRQlNPYJyLuznF+R3oGC3P4DEnrDUS2NxXJ44PnSDN9tUhagfQx4BmS4l6l6L8rkGbxCl6Zr0u0UUTcRVKMILdRb4mIZyNiPmmmtIlPk5THb+XnX5Z/kZaSvCp5P19u5xdp9RWAOyLioIj4R44zBzgvh60MbNpUgNJS0dmPPvpoUzRjjFmmsWJnjDE1qHUg+edrgsvGIhqPPsgD8IILssJWDp8LnFHy2rMTspnCXP4PI+L+klyU4hxGmvW4PiJeaKhGMTBeETioUoargf/Lfy7VDpK2Iy3bvLwSdGC+np/3Ci7hgK1LcbdoKFeviYh/kvZHXpVdmfml+/GVsPIz+O+sSJTTvYe0bBbSrOBbOiTbG55qE/ZOkiJ7Y0TMqwn/IOnZ717uGyRF/XrgghqZop2qbdRbassraQItQ0S/rosT6czIWfnPfSStWwouHwsxi6UpK5SN+zYj4ryImB4R09dYY42maMYYs0xjxc4YY+o5CvhzRFxTE/ZD4N58v5+k1QeQzw9K96/ooGwxE9aksAG8I18faRNnVun+jTXhX87XnSRtXgk7AvhqVakBXpOvbycpcVX3UtKS0rVJS0kHTEQcGRFvybM4hVGR04Cfl6L1539it55fW6oznRWK5/RAg+yTEXFuRPy84n9+RLw2Iq4HkLSKpKMk3Uh6FtD/cUNTeXeipXD1ph+OAl5f8u/pWIgnS/f9VUqNMWZYYMXOGGMqZAt8BwGbNswo/YO0FwzSXp8P9DeviPg7rSV5fTobbyCyklYizaZB86CbiLiP1mzL2jVRLgIez/cfKqW/OvA2apZ40mq7RRHxcA/umT5Uq0ck7SDpCtJ+rFtzGQfC/5Xu+3q24UBk21HMaI3tj7CkqZL+m2Q4aEXSMtt/dKhsVTYu3bdTVm8v3df1Q2OMWe6xYmeMMUtzCOlL/ybUzyhtTbIWWCgdh0vq1yA6M69yHQzZlUr3Pc04FmlX96eRl4gWSwrfJ2mVfH8waRnonJr0irZaan9it5A0WdKFwE+Bb0fEthFxIclq5ECY13Dfbdl2FP/bN+yLUD5+4liSEjUX2CIiTo2IdjNpA2Vi6b5dP5xbul+qHxpjjLFiZ4wxS6DWgeRfiYgH28wk3QlcnMXWIZ1311+KJWJ3DaLsHFpKTXUJZRXl618bws8mLYlbETg4nxl2GPCVNnlDOoahOVNp7bzXcUBkYyJXAf8GvCciLu5BpC+Ul/f19RkMRLYdhSK2laTGDWOSRlfOs/sC6ViGsyLi+GwEptuU9/i164cq3Tf1Q2OMWa6xYmeMMUvyTtLMwXk9RaS1vwzSnrx21P7e5mWfq+Y/r66LM0DZWrKxlOvynxtIenlDHqNLeXy/Ia37aO0XO4K0d+6hiJjdkP3v83VXSTu0KeZRtDF40QcOIh0x8HhEXNHPNJr+X5ZnxX7ZEGcgsv3hd/k6BjiyTbwZ5GWNkv6FlgGcCztYlp6YRWsJ5u5t4hUK6kMkAy/GGGMqWLEzxpglOQ74TkQ83lPEiPgTLSVlux6UlFUb/HclzUbcTvPgfiCy7X7nzyrdH9EQZ3PSuWl30l55/FK+vgz4b5pn66A10yngu5I2rkaQNA3YOSJuaZNObylmgiZJWrESVl5CO7pNGk3PoLBm+fN8zESnZNVwnzykduGX0FomfFy2TlqV34x01EWh6JZny9asxBVJSYT6NioM9Cxe3itpbP4oUC3jEmXNRxNclv/cPj/3Ogr/r1YMx/RlHLNUOxpjzEjCip0xxmQk7Uc6gPwnfRD7Wen+lMqAu8y/5mWe5fzGkQ4NX0Q6IDxqJfsnu0K+Np6PlmevLs1/HiZpek20I0hniB3S5kgEIuI3wM3FnyTloonLaFk5XBe4SdKnJG0n6XWSTiApqse3SaMv3JevY4GvSBqfFY/3suRRDOtJ2kTSYTVp7F/1yPsJjyYZlzm6Tf79kS0bU5lUE16eyZxcDoiIR4FT8p8TgGty+24r6Q2STgJuAk4vGae5r5TEaZLWyuXcAfgVrRmz9SStlNMoKD6CvCbvZVwRmEn6IFAt7xJlzRxTSuPs6n5VSVOA/UjGbj5bkV2tdL8C7anL2xhjRg7R5RPQ7ezs7JZ1RxoQ7kMyyhCkGaUNALWRGQ1MBa7JMoU7n2Tpb3SOV/jPIymMryIZjNiStO/rBZLSVJdHn2Vz+J6kc9QCeB54HzC5Td1/nuM+RlJCppAUrtNJVjff0ct2PCCn8+lexF2DdDh41LhFwJEdfL4vBRaU0n+apFDdQJqpWpT9nwNuA1bPclMrz+AC0rEEE0lHNszO6exZk2e/ZEmzSquTjnko96nVi/5Imhk7qRT+R9KyzjGldEYB32ho3wBOrunPN5fCXyQZLHmQdCRB+VnNAV5bkv1Apa5zgYNy2Hhgj9xXg2SU6I3A+Er+2wGP5jjXANNzW21HmhW/CVi3IrNqbpsi78tJll5H5XZcjaRgFuFXkmYjG9/riGDatGlhjDGdBpgd3R7PdDsDOzs7u2XdAT9uGPye0UbmY20GzQHMyPGKvz9AmkW5h6RAPAL8D7BNmzz6LEsywlFbpjb5iHRo+CyScreQZIr/TGBqH9pxHMkYxjp9iH8McAtJyZlLUjJ36sIzfg1pb9ZC0qHV/0FWhEjGXxaQ9gmuVZKZWmq/vXO8f+Rn8ABJcdqoIb9+yQI7tOlTe5GWRDaFn1aT3r7Ab3L9niR9ENiloczrkvZRPkFSss6jpeS+j6S03QhsV5EbDXwxh98N/Hsp7DcNZb2hJv81SDNyt+X+8E/SoeUHAxMqcTdr0w5HkvYPNrZju75ixc4Y0w0YBMWu+PpnjDGmC0gqfmT/PSJmDpasGTiSppKUaUjK5qzBkDVDy/Tp02P27Ca7P8YY0z8k3RQRdVseOob32BljjDHGGGPMMMeKnTHGGGOMMcYMc6zYGWNMl5A0sfTnSo0ROyxrOsZAnoGfnzHGmEHFip0xxnQYJVYhGVgpmCFpo6op907Kms6Qn8GatA7sBvigpHUrZ7N1VNYYY4wZCFbsjDGm87yTdC7XiSW/aSSLld/qouyIRNKcAbjj+pHlR4GHgUNLfnuQrFl+pouyxhhjTL+xVUxjjDHLNMVh2f1kQUQs6FhhzIjHVjGNMd1gMKxijulm4sYYY8xAiYiHh7oMxhhjzLKOl2IaY4wxxhhjzDDHip0xxhhjjDHGDHOs2BljjDHGGGPMMMfGU4wxxhhjMpLmA3cMdTm6yOrAnKEuRBcZyfUbyXWDkV+/l0fEpG5mYOMpxhhjjDEt7ui25bqhRNJs1294MpLrBstH/bqdh5diGmOMMcYYY8wwx4qdMcYYY4wxxgxzrNgZY4wxxrQ4b6gL0GVcv+HLSK4buH4DxsZTjDHGGGOMMWaY4xk7Y4wxxhhjjBnmWLEzxhhjjDHGmGGOFTtjjDHGjCgkjZZ0sKSbJC2UdK+kMyRNHmC660g6O6e3UNINkt7TqXL3oRxdqV9Oe5SkfSXdLGlGB4rbnzJ0vH6StpL0bUkPSXpO0oOSvitp0M3rd6l+60n6mqQHSvW7UNKGnSx7L8rRtb5ZyWd9SXMl3dvJdHuRbzffvd9IigY3o1dpeI+dMcYYY0YKklYCLgdeAxwJ/ADYFLiI9EF7p4i4rx/p7pDTvQ84BLgL2A/4CvAN4NAYhEFVF+s3HjgAOBbYJHv/e0TM7ECx+1KOjtdP0nuBmcC4muAXgaMj4qwBFLsvZelG/TYB/hdYoyb4cWCHiLhtIOXuZTm60jdr8hkF/BLYEfh7REwdaJq9zLdr9ZP0euC6huC5wDoR8XSPCUWEnZ2dnZ2dnd2IcHngFcDBFf/NgUXArcCEPqa5YR5cPQa8pBJ2es7vhOFavyx/JPBu0iA1spsx3J8f8ErgOeAGYH9gGrAL8K1SPRcBew7T+k0A/kr6wPBKYDKwPUnxKer3y+FYtzb5fKxUt3uHa9+spHEF8Ahwe437dK/TGazGsLOzs7Ozs7PrpgPemwde9wOja8J/nMNP7WO6P89yp9SErZUVh+eAzYdj/SpprD1Uil036gdcDHy1IeyYUl1vGab1OxZ4f43/WOCmnN6LwMThVreGfLYEFgA/HUzFrpv1y3V6AdhsoOX0HjtjjDHGjBQ+la8/iYgXa8J/kK9HSprUmwQlbQfsmv+8vBoeEQ+TZoPGkgbZ3aTj9avhsX7KdYKO1k/SaNKy0iPqwiPi88Dv859bd3ofWA3deH5XRcQ3qp4R8Txwbv5TfStmv+h638zLhS8mzZL/rj9pDIBu1u944NKIuL3fpctYsTPGGGPMsEfStsAr8p83NUS7MV9XAv61l0kfmK8vAn/oId198z6cjtPF+lV5vp9yA6JL9VsJ+GhELGoTp6ys1+3B6wjden4R8cc2wffn67URsbA36fWHQeybpwLzgc/0U75fdLN+kjYA9gUelfQaSWP7XVCs2BljjDFmZLBr6f6ehjh3lu5f38d0H4qIZxvi3JGvE4BX9zLdvtKt+i1B5LVhQ0DH6xcR8yLi1z1Em5Ov8yNiTtuYA2NQnl+FV5AU9eM6kFY7ul43SW8CDgLe1zBj1k26Wb/jgNGkPa6/BR6U9CVJa/etiAkrdsYYY4wZCWxdur+/LkJWzOblP1/VU4J59m2jdmlm/lm67zHdftLx+i1jDFX9igH0jzqUXhODWj9JK5Kstx4SETf2FH+AdLVuklYhWTU9KiL+1p8CDpCu1E/SmsCMivfqwIeBv0o6qG/FtGJnjDHGmJHB1NJ9u31iT+Xrml1Is7fp9oeppftO1W9ZYmrpfjDrt1O+nt2h9JqYWrrvav0krU7a8/UkyfBPt5lauu9G3b4KXB8R3+yjXKeYWrrvZP2eJFnU3JZ0dMp5pKWmABOBb0j6RO+LacXOGGOMMSODsuGLpxpjJetzAFO6kGZv0+0P3ajfssSg1y+f//Z6kuGK6weaXg90vX6SXi3pVNLxB7uSFIa7Jb27r2n1ka7VTdL7gNcBH+hHuTpFV+oXEU9HxN0RcWNEfCciPkA6F+/8UrSTJe3akMRSWLEzxhhjzEigbPmvaS8cJOuVkEyTdzrN3qbbH7pRv2WJoajfCaQDvD/UgbR6YjDqtwi4Bfg2rWWBKwLfyQdgd4uu1E3SS4EvAwdGxBP9LFsnGLS+GREPR8T7SZYyC07urbwVO2OMMcaMBOaX7ttZN5yQr092Ic3eptsfulG/ZYlBrZ+kXYD3APvnIyu6TdfrFxE3RcT3IuIIYAPguzloNH1QDvpBx+smaRRwIfCNiLhmAGXrBIP+7kXEZ0lLUAFe3VtjKlbsjDHGGDMSKBs1qD1HKg8WV66J3+80M6s2yHSSbtRvWWLQ6idpLZLScHREXNXfdPrIoD6/PMO1P+mMRYDXShozkDTb0I26fQxYH/iqpPWqjtbyyNEl/1X6W4EeGKp37z9pzRC+tDcCVuyMMcYYMxIonzG3bkOcl9BaVtXu/C9g8eD4vh7ShCWNJfSYbj/peP2WMQalfpImAJcCMyPirP6k0U8G/fnl8/vOzH+OIVlc7AbdqNsHgQ2Bu0mKUtUdneOtV/L7Qp9K3XuG5N3Lx28Uez8X9EbGip0xxhhjRgJl63+vaIizUem+t8u7inQ3kTS6h3QXAr/vZbp9pVv1W1boev3yrMq3gT9HxMf7Kj9Ahur5zcrX50n7CbtBN+qmnqMMGkP57j1E+l3p1TEPVuyMMcYYMxK4Hrgr32/fEGd6vs4HftjLdL+VrysCW/aQ7qUR8XQv0+0r3arfskJX6ydJwNdIVg2P6E8BB8hQPb9ipuc3EfFch9Ks0vG6RcTUiFCTA07KUf9e8p8xkEq0YSjfvbWBS3r7u2LFzhhjjDHDnogIWgYi9s4D+Spvy9cvR8T8mvC6dH9Na9Zjn2q4pJeRFL4XgM/2pcx9oVv1q1JJd9BmTbpZv5JStwowIy9RrIt3gKSX96HYvWawnl8Nr83X0zuU3lIMYd0GhaGqn6T1gS2AT/VWxoqdMcYYY0YKFwJXkQwNvKscIGk70mHUtwOnVML2kvRPSbdK2qAm3UNJy6EOlbRSJez/kcZTn4mI2zpTjUa6Vb8yE0v3Kw68yH2i4/XLy2fPB7YmHW+wsaTNSm4LSW+SdDZpAH1ntypHd+q3paQ3S1rKWmPeT/hZ4EsR8bPOVmUpBqNvDiXdeHa7SDpcUtn4UhG2AuljxP4R8UCvSxkRdnZ2dnZ2dnYjwgFrADeR9hPtRbJU93bSXpVbgZfVyPyYdPZUAMc0pLsXaRnfL4BNgLWAU7PM2YCGef1GA6uRlJsi7q+BjYHxw/H5AeOBy0phPbmThln9JpH2zgXJyM+HgJeRlPM3AtcBnxyOz64XeZ2YZe4drvUDHsn+/7+9uw/Vs67jOP7+nDVxD5ljpYUOBUv/mU+EpX+ElSupBc1qsCyirAWmlZlOrX96IEtbolEQKklRLNcIQhzSA0iFIixJ0sgicmupOaVR00zNb39c19y1u3Ofc59zhnHd9/sFP7geftf3+t2H88f5nN91X7+/ARfTvJjlCGAtzVIVr5/zGF+sH4bNZrPZbDbbi9GA5e0ffn8Anqb5T/pngCVD+q8F9gAPAMfPUPdk4IfAYzQzeD8Fzh2HzwdsniHwPNTHz0czyzJqqCvghD59vvbcpzp1nm1Dwl3AF4BV4/C7OeS6z/X5d7M99waaF608QbOswS6a7+edDyyez/jSFpYkSZIk9ZTfsZMkSZKknjPYSZIkSVLPGewkSZIkqecMdpIkSZLUcwY7SZIkSeo5g50kSZIk9ZzBTpIkSZJ6zmAnSZIkST1nsJMkSZKknjPYSZIkSVLPGewkSZLUW0muT1LTtHPmUWsqyYND6h050Pe8JNuT7Eny7yS72/2LkhyX5C1JNg9cc1WSLUn2DbnHTO1LC/1ZabwZ7CRJktRnlwPHAV8ZOH7VPGqtA07s7N8LnAYcXlV7AZIsTnIr8CPgEWAN8HLgHOBXwOeBh4CfDBavqi9X1XuBtZ3Du4FV07TjgFOATwBPAi+dx+fRBElV/b/HIEmSJC1Ykt3AMZ1DZ1TVjjlcfzdwZufQhqq6daDPDTRha3NVXT5NjWOB7cDJwNeq6rJp+iwD9rW7O6vq+FnGdSHwuqr60KifRZPHGTtJkiSNi53APzv7I8/aJTmbJtT9o3N4z0CfVwIXtbvXT1enqnYD7wb+NcPtnhp1XK0twNI5XqMJY7CTJEnSuHgW+C4HgtO6JCeNeO0m4K/A7Z1jzw/0OQtY1G4Pfeytqv4IbJ3h/JwemWsfA71mLtdo8hjsJEmSNE4eB77dbk8BV8x2QZLVwNuA64BnZura2f7iLGW3zXbfuaiqew9lPY0fg50kSZLGzWbguXb7/e333mayCdgL3DhLvx0cmMW7IMnWJCuG9P31SCOdRZIfHIo6Gn8GO0mSJI2VqtoJ7H/pyWLg08P6JlkFbAC+WVX7hvVr6+4CvtM5tB54MMmHk0wN9H1kuhenzEWSxcCoj5JqwhnsJEmSNI6u4cD34DYmWTmk36U0s3tfH7Hux4Gfd/ZfAdwM3JfkHfMZ6HSSLAWuBobNCEoHMdhJkiRp7FTVb2mWHQBYRhPIDtI+RvkR4Jaq2jN4fkjdJ4G306yb91zn1GrgtiR3JjltDkNdleTRgfYYzds5FzTjp8lisJMkSdK46r5J8uJ2/biujwFLaL6TN7KqeqaqrgJO5+DZO4CzgR1Jrh58PHOIh2kWQe+21wJvBn42l3FpshnsJEmSNJaq6pfAXe3uSuCj+88lOZxmofGtVfXneda/v6rWAGuB+zunFtGsoXfLCGX+U1WPDrS/VNUvgPOYeT086QUGO0mSJI2z7qzdpe0LSQA+CBwFXLvQG1TVduBUmuD4eOfUB5KsW0DdfcB9CxyeJoTBTpIkSePsNuCBdvtYmuUPpmjelHlHVf1m1EJJVie5crpzVfV8Vd1EE/Ae6Jy6YH7DfqHuhoVcr8lhsJMkSdLYqqri4Fm5K2iWKXg1B8/mjeqsWe73MPAeDqx395p53GNaSTJ7L00qg50kSZLG3RZgV7t9EvAt4J6qunMetc5JsnymDlX1ew48Qvn3edxjmO8lOfIQ1tMYMdhJkiRpXEwB/zOrVVXPAtd1Dh3J8Nm6qSHb+y0DNo0wlv3h7/bBE/OZeUvyRmB5Ve2d67WaDAY7SZIkjYuj2zadm4An2u0HgR8P6XfUkO2uzybZOGwQSdbTPIL5J+CGabos6WwvGlanU+9VwPeBrbP11eQy2EmSJKm3kkwlWZnkEuBE4F1J1iRZ1p0Zq6qngG+0u9dW1fOdGknysiTnA2/qlL+kfWHKYQO3nQJuTHJHkncmOSbJEUlOTXIdTQj7HfDW9s2Wg07vbB+d5MwkhyV5SactTXJsO6Z7gBUMD6MSab5PKkmSJPVPkuuBTw45vb6qtnX6rgTuBlZX1TOd45cBX53lVito3qp5IXAlcAZwLk0QPAE4AthHs57drcDNVfX0wFg3AqcA72vrzcW2qlo/x2s0QQx2kiRJktRzPoopSZIkST1nsJMkSZKknjPYSZIkSVLPGewkSZIkqecMdpIkSZLUcwY7SZIkSeo5g50kSZIk9ZzBTpIkSZJ6zmAnSZIkST1nsJMkSZKknjPYSZIkSVLPGewkSZIkqef+CwCrkeH+QVgSAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<Figure size 360x576 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "mses = df_plot.rename(columns=inv_aliases).loc[875].sort_values(ascending=True)\n",
    "mses.tail(10).plot.barh(figsize=(5,8), colormap='PRGn')\n",
    "\n",
    "ax = plt.gca()\n",
    "for item in ([ax.title, ax.xaxis.label, ax.yaxis.label] +\n",
    "                 ax.get_xticklabels() + ax.get_yticklabels()):\n",
    "    item.set_fontsize(30)\n",
    "\n",
    "ax.set_xlim((0, 0.5))\n",
    "ax.set_xlabel('MSE')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 184,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Text(0.5, 0, 'MSE')"
      ]
     },
     "execution_count": 184,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 360x576 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "mses = df_plot.rename(columns=inv_aliases).loc[7125].sort_values(ascending=True)\n",
    "mses.tail(10).plot.barh(figsize=(5,8), colormap=\"PRGn\")\n",
    "\n",
    "ax = plt.gca()\n",
    "for item in ([ax.title, ax.xaxis.label, ax.yaxis.label] +\n",
    "                 ax.get_xticklabels() + ax.get_yticklabels()):\n",
    "    item.set_fontsize(30)\n",
    "\n",
    "ax.set_xlim((0, 0.5))\n",
    "ax.set_xlabel('MSE')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 81,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'Appraise_property'"
      ]
     },
     "execution_count": 81,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "inv_aliases[\"A\"]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'START': 'Loan__application_received',\n",
       " 'A': 'Appraise_property',\n",
       " 'B': 'Approve_application',\n",
       " 'C': 'Assess_eligibility',\n",
       " 'D': 'Assess_loan_risk',\n",
       " 'E': 'Cancel_application',\n",
       " 'F': 'Check__application__form_completeness',\n",
       " 'G': 'Check_credit_history',\n",
       " 'H': 'Check_if_home_insurance_quote_is_requested',\n",
       " 'I': 'Prepare_acceptance_pack',\n",
       " 'J': 'Receive_updated_application',\n",
       " 'K': 'Reject_application',\n",
       " 'L': 'Return_application_back_to_applicant',\n",
       " 'M': 'Send_acceptance_pack',\n",
       " 'N': 'Send_home_insurance_quote',\n",
       " 'O': 'Verify_repayment_agreement',\n",
       " 'END_A': 'Loan__application_approved',\n",
       " 'END_R': 'Loan_application_rejected',\n",
       " 'END_C': 'Loan__application_canceled'}"
      ]
     },
     "execution_count": 40,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "inv_aliases"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 63,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1296x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "df_plot[[\"G\", \"A\", \"L\"]].rename(columns=inv_aliases).plot(figsize=(18,4), lw=4, color = ['#35b588', 'y', '#ff5f54'])\n",
    "plts.plot_drift_vertical_lines(c['log_size'], label=\"True drift\")\n",
    "\n",
    "ax = plt.gca()\n",
    "for item in ([ax.title, ax.xaxis.label, ax.yaxis.label] +\n",
    "                 ax.get_xticklabels() + ax.get_yticklabels()):\n",
    "    item.set_fontsize(32)\n",
    "\n",
    "# ax.set_xlabel('índice dos $\\it{traces}$')\n",
    "ax.set_xlabel(None)\n",
    "# ax.set_xlim((0, df.index.max() + 100))\n",
    "# bbox_to_anchor=(0.5, 1.3),\n",
    "leg = plt.legend(fontsize=32, loc='upper center', bbox_to_anchor=(0.5, -0.1),\n",
    "          fancybox=True, shadow=False, prop={\"family\":\"Times New Roman\", \"size\":\"32\"},\n",
    "          frameon=False, ncol=2, labelspacing=0.1, columnspacing=1)\n",
    "\n",
    "for line in leg.get_lines():\n",
    "    line.set_linewidth(5)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 249,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>log_size</th>\n",
       "      <th>model</th>\n",
       "      <th>representation</th>\n",
       "      <th>sliding_window</th>\n",
       "      <th>tipo_mudanca</th>\n",
       "      <th>window_size</th>\n",
       "      <th>y_true</th>\n",
       "      <th>localizations</th>\n",
       "      <th>roc_scores</th>\n",
       "      <th>avg_precisions</th>\n",
       "      <th>AVG_roc_scores</th>\n",
       "      <th>AVG_avg_precisions</th>\n",
       "      <th>flag_binary_support</th>\n",
       "      <th>flag_frequency_support</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>1241</th>\n",
       "      <td>5000</td>\n",
       "      <td>kmeans__k=2</td>\n",
       "      <td>binary</td>\n",
       "      <td>False</td>\n",
       "      <td>cb</td>\n",
       "      <td>250</td>\n",
       "      <td>[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 0, 0, 0, ...</td>\n",
       "      <td>[{'H': 0.15849869354777202, 'I': 0.15849869354...</td>\n",
       "      <td>[1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0]</td>\n",
       "      <td>[1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0]</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "      log_size        model representation  sliding_window tipo_mudanca  \\\n",
       "1241      5000  kmeans__k=2         binary           False           cb   \n",
       "\n",
       "      window_size                                             y_true  \\\n",
       "1241          250  [0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 0, 0, 0, ...   \n",
       "\n",
       "                                          localizations  \\\n",
       "1241  [{'H': 0.15849869354777202, 'I': 0.15849869354...   \n",
       "\n",
       "                                         roc_scores  \\\n",
       "1241  [1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0]   \n",
       "\n",
       "                                     avg_precisions  AVG_roc_scores  \\\n",
       "1241  [1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0]             1.0   \n",
       "\n",
       "      AVG_avg_precisions  flag_binary_support  flag_frequency_support  \n",
       "1241                 1.0                 True                    True  "
      ]
     },
     "execution_count": 249,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "resp_df[\n",
    "    (resp_df.tipo_mudanca == \"cb\") &\n",
    "    (resp_df.representation == 'binary') &\n",
    "    (resp_df.model == \"kmeans__k=2\") &\n",
    "    (resp_df.window_size == 250) & \n",
    "    (resp_df.log_size == 5000)\n",
    "]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Versão com todas as janelas anteriores"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 285,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>log_size</th>\n",
       "      <th>model</th>\n",
       "      <th>representation</th>\n",
       "      <th>sliding_window</th>\n",
       "      <th>tipo_mudanca</th>\n",
       "      <th>window_size</th>\n",
       "      <th>y_true</th>\n",
       "      <th>localizations</th>\n",
       "      <th>roc_scores</th>\n",
       "      <th>avg_precisions</th>\n",
       "      <th>AVG_roc_scores</th>\n",
       "      <th>AVG_avg_precisions</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>10000</td>\n",
       "      <td>kmeans__k=2</td>\n",
       "      <td>transitions</td>\n",
       "      <td>False</td>\n",
       "      <td>cf</td>\n",
       "      <td>75</td>\n",
       "      <td>[0, 0, 0, 1, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 1, ...</td>\n",
       "      <td>[{'M N': 0.22222222444444445, 'M O': 0.0665006...</td>\n",
       "      <td>[0.7916666666666667, 0.875, 0.875, 0.75, 0.875...</td>\n",
       "      <td>[0.7794871794871795, 0.8678266178266179, 0.867...</td>\n",
       "      <td>0.787037</td>\n",
       "      <td>0.795094</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>10000</td>\n",
       "      <td>kmeans__k=2</td>\n",
       "      <td>transitions</td>\n",
       "      <td>False</td>\n",
       "      <td>cf</td>\n",
       "      <td>100</td>\n",
       "      <td>[0, 0, 0, 1, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 1, ...</td>\n",
       "      <td>[{'M N': 0.5, 'M O': 0.1408642408780667, 'N O'...</td>\n",
       "      <td>[0.75, 0.875, 0.875, 0.7083333333333334, 0.791...</td>\n",
       "      <td>[0.7589476495726496, 0.857905982905983, 0.8678...</td>\n",
       "      <td>0.777778</td>\n",
       "      <td>0.790776</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>10000</td>\n",
       "      <td>kmeans__k=2</td>\n",
       "      <td>transitions</td>\n",
       "      <td>False</td>\n",
       "      <td>cf</td>\n",
       "      <td>125</td>\n",
       "      <td>[0, 0, 0, 1, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 1, ...</td>\n",
       "      <td>[{'M N': 0.5, 'N O': 0.13857436956133617, 'M O...</td>\n",
       "      <td>[0.75, 0.75, 0.875, 0.7083333333333333, 0.7916...</td>\n",
       "      <td>[0.7589476495726496, 0.7606837606837606, 0.867...</td>\n",
       "      <td>0.763889</td>\n",
       "      <td>0.779486</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>10000</td>\n",
       "      <td>kmeans__k=2</td>\n",
       "      <td>transitions</td>\n",
       "      <td>False</td>\n",
       "      <td>cf</td>\n",
       "      <td>150</td>\n",
       "      <td>[0, 0, 0, 1, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 1, ...</td>\n",
       "      <td>[{'M N': 0.5, 'N O': 0.140847231039704, 'M O':...</td>\n",
       "      <td>[0.7083333333333333, 0.8333333333333333, 0.875...</td>\n",
       "      <td>[0.7446018893387314, 0.8113782051282051, 0.867...</td>\n",
       "      <td>0.768519</td>\n",
       "      <td>0.781490</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>10000</td>\n",
       "      <td>kmeans__k=2</td>\n",
       "      <td>transitions</td>\n",
       "      <td>False</td>\n",
       "      <td>cf</td>\n",
       "      <td>200</td>\n",
       "      <td>[0, 0, 0, 1, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 1, ...</td>\n",
       "      <td>[{'M N': 0.5, 'N O': 0.13391271574047678, 'M O...</td>\n",
       "      <td>[0.75, 0.7083333333333333, 0.875, 0.7083333333...</td>\n",
       "      <td>[0.7606837606837606, 0.7467948717948718, 0.857...</td>\n",
       "      <td>0.763889</td>\n",
       "      <td>0.778891</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1291</th>\n",
       "      <td>7500</td>\n",
       "      <td>kmeans__k=3</td>\n",
       "      <td>binary</td>\n",
       "      <td>False</td>\n",
       "      <td>lp</td>\n",
       "      <td>100</td>\n",
       "      <td>[1, 0, 1, 1, 0, 1, 1, 1, 0, 1, 0, 0, 0, 0, 0, ...</td>\n",
       "      <td>[{'E': 0.12244897959183675, 'END_A': 0.1224489...</td>\n",
       "      <td>[0.3977272727272727, 0.3977272727272727, 0.397...</td>\n",
       "      <td>[0.4407894736842105, 0.4407894736842105, 0.440...</td>\n",
       "      <td>0.392677</td>\n",
       "      <td>0.436160</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1292</th>\n",
       "      <td>7500</td>\n",
       "      <td>kmeans__k=3</td>\n",
       "      <td>binary</td>\n",
       "      <td>False</td>\n",
       "      <td>lp</td>\n",
       "      <td>125</td>\n",
       "      <td>[1, 0, 1, 1, 0, 1, 1, 1, 0, 1, 0, 0, 0, 0, 0, ...</td>\n",
       "      <td>[{'N': 0.036978619968428945, 'M': 0.0369786199...</td>\n",
       "      <td>[0.3522727272727273, 0.3977272727272727, 0.397...</td>\n",
       "      <td>[0.3991228070175438, 0.4407894736842105, 0.440...</td>\n",
       "      <td>0.392677</td>\n",
       "      <td>0.436160</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1293</th>\n",
       "      <td>7500</td>\n",
       "      <td>kmeans__k=3</td>\n",
       "      <td>binary</td>\n",
       "      <td>False</td>\n",
       "      <td>lp</td>\n",
       "      <td>150</td>\n",
       "      <td>[1, 0, 1, 1, 0, 1, 1, 1, 0, 1, 0, 0, 0, 0, 0, ...</td>\n",
       "      <td>[{'E': 0.16666666666666666, 'END_A': 0.1666666...</td>\n",
       "      <td>[0.3977272727272727, 0.3977272727272727, 0.397...</td>\n",
       "      <td>[0.4407894736842105, 0.4407894736842105, 0.440...</td>\n",
       "      <td>0.397727</td>\n",
       "      <td>0.440789</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1294</th>\n",
       "      <td>7500</td>\n",
       "      <td>kmeans__k=3</td>\n",
       "      <td>binary</td>\n",
       "      <td>False</td>\n",
       "      <td>lp</td>\n",
       "      <td>200</td>\n",
       "      <td>[1, 0, 1, 1, 0, 1, 1, 1, 0, 1, 0, 0, 0, 0, 0, ...</td>\n",
       "      <td>[{'E': 0.07407407407407408, 'END_A': 0.0740740...</td>\n",
       "      <td>[0.3977272727272727, 0.3977272727272727, 0.318...</td>\n",
       "      <td>[0.4407894736842105, 0.4407894736842105, 0.421...</td>\n",
       "      <td>0.388889</td>\n",
       "      <td>0.438596</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1295</th>\n",
       "      <td>7500</td>\n",
       "      <td>kmeans__k=3</td>\n",
       "      <td>binary</td>\n",
       "      <td>False</td>\n",
       "      <td>lp</td>\n",
       "      <td>250</td>\n",
       "      <td>[1, 0, 1, 1, 0, 1, 1, 1, 0, 1, 0, 0, 0, 0, 0, ...</td>\n",
       "      <td>[{'E': 0.16666666666666666, 'END_A': 0.1666666...</td>\n",
       "      <td>[0.3977272727272727, 0.3181818181818182, 0.397...</td>\n",
       "      <td>[0.4407894736842105, 0.42105263157894735, 0.44...</td>\n",
       "      <td>0.371212</td>\n",
       "      <td>0.434211</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>1296 rows × 12 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "      log_size        model representation  sliding_window tipo_mudanca  \\\n",
       "0        10000  kmeans__k=2    transitions           False           cf   \n",
       "1        10000  kmeans__k=2    transitions           False           cf   \n",
       "2        10000  kmeans__k=2    transitions           False           cf   \n",
       "3        10000  kmeans__k=2    transitions           False           cf   \n",
       "4        10000  kmeans__k=2    transitions           False           cf   \n",
       "...        ...          ...            ...             ...          ...   \n",
       "1291      7500  kmeans__k=3         binary           False           lp   \n",
       "1292      7500  kmeans__k=3         binary           False           lp   \n",
       "1293      7500  kmeans__k=3         binary           False           lp   \n",
       "1294      7500  kmeans__k=3         binary           False           lp   \n",
       "1295      7500  kmeans__k=3         binary           False           lp   \n",
       "\n",
       "      window_size                                             y_true  \\\n",
       "0              75  [0, 0, 0, 1, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 1, ...   \n",
       "1             100  [0, 0, 0, 1, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 1, ...   \n",
       "2             125  [0, 0, 0, 1, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 1, ...   \n",
       "3             150  [0, 0, 0, 1, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 1, ...   \n",
       "4             200  [0, 0, 0, 1, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 1, ...   \n",
       "...           ...                                                ...   \n",
       "1291          100  [1, 0, 1, 1, 0, 1, 1, 1, 0, 1, 0, 0, 0, 0, 0, ...   \n",
       "1292          125  [1, 0, 1, 1, 0, 1, 1, 1, 0, 1, 0, 0, 0, 0, 0, ...   \n",
       "1293          150  [1, 0, 1, 1, 0, 1, 1, 1, 0, 1, 0, 0, 0, 0, 0, ...   \n",
       "1294          200  [1, 0, 1, 1, 0, 1, 1, 1, 0, 1, 0, 0, 0, 0, 0, ...   \n",
       "1295          250  [1, 0, 1, 1, 0, 1, 1, 1, 0, 1, 0, 0, 0, 0, 0, ...   \n",
       "\n",
       "                                          localizations  \\\n",
       "0     [{'M N': 0.22222222444444445, 'M O': 0.0665006...   \n",
       "1     [{'M N': 0.5, 'M O': 0.1408642408780667, 'N O'...   \n",
       "2     [{'M N': 0.5, 'N O': 0.13857436956133617, 'M O...   \n",
       "3     [{'M N': 0.5, 'N O': 0.140847231039704, 'M O':...   \n",
       "4     [{'M N': 0.5, 'N O': 0.13391271574047678, 'M O...   \n",
       "...                                                 ...   \n",
       "1291  [{'E': 0.12244897959183675, 'END_A': 0.1224489...   \n",
       "1292  [{'N': 0.036978619968428945, 'M': 0.0369786199...   \n",
       "1293  [{'E': 0.16666666666666666, 'END_A': 0.1666666...   \n",
       "1294  [{'E': 0.07407407407407408, 'END_A': 0.0740740...   \n",
       "1295  [{'E': 0.16666666666666666, 'END_A': 0.1666666...   \n",
       "\n",
       "                                             roc_scores  \\\n",
       "0     [0.7916666666666667, 0.875, 0.875, 0.75, 0.875...   \n",
       "1     [0.75, 0.875, 0.875, 0.7083333333333334, 0.791...   \n",
       "2     [0.75, 0.75, 0.875, 0.7083333333333333, 0.7916...   \n",
       "3     [0.7083333333333333, 0.8333333333333333, 0.875...   \n",
       "4     [0.75, 0.7083333333333333, 0.875, 0.7083333333...   \n",
       "...                                                 ...   \n",
       "1291  [0.3977272727272727, 0.3977272727272727, 0.397...   \n",
       "1292  [0.3522727272727273, 0.3977272727272727, 0.397...   \n",
       "1293  [0.3977272727272727, 0.3977272727272727, 0.397...   \n",
       "1294  [0.3977272727272727, 0.3977272727272727, 0.318...   \n",
       "1295  [0.3977272727272727, 0.3181818181818182, 0.397...   \n",
       "\n",
       "                                         avg_precisions  AVG_roc_scores  \\\n",
       "0     [0.7794871794871795, 0.8678266178266179, 0.867...        0.787037   \n",
       "1     [0.7589476495726496, 0.857905982905983, 0.8678...        0.777778   \n",
       "2     [0.7589476495726496, 0.7606837606837606, 0.867...        0.763889   \n",
       "3     [0.7446018893387314, 0.8113782051282051, 0.867...        0.768519   \n",
       "4     [0.7606837606837606, 0.7467948717948718, 0.857...        0.763889   \n",
       "...                                                 ...             ...   \n",
       "1291  [0.4407894736842105, 0.4407894736842105, 0.440...        0.392677   \n",
       "1292  [0.3991228070175438, 0.4407894736842105, 0.440...        0.392677   \n",
       "1293  [0.4407894736842105, 0.4407894736842105, 0.440...        0.397727   \n",
       "1294  [0.4407894736842105, 0.4407894736842105, 0.421...        0.388889   \n",
       "1295  [0.4407894736842105, 0.42105263157894735, 0.44...        0.371212   \n",
       "\n",
       "      AVG_avg_precisions  \n",
       "0               0.795094  \n",
       "1               0.790776  \n",
       "2               0.779486  \n",
       "3               0.781490  \n",
       "4               0.778891  \n",
       "...                  ...  \n",
       "1291            0.436160  \n",
       "1292            0.436160  \n",
       "1293            0.440789  \n",
       "1294            0.438596  \n",
       "1295            0.434211  \n",
       "\n",
       "[1296 rows x 12 columns]"
      ]
     },
     "execution_count": 285,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "resp__2 = pd.DataFrame.from_dict(resp__2)\n",
    "resp__2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 289,
   "metadata": {},
   "outputs": [],
   "source": [
    "resp__2[\"flag_binary_support\"] = resp__2[\"tipo_mudanca\"].isin(binary_support + ['re'])\n",
    "resp__2[\"flag_frequency_support\"] = resp__2[\"tipo_mudanca\"].isin(binary_support + ['re', 'lp'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 290,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>model</th>\n",
       "      <th>representation</th>\n",
       "      <th>sliding_window</th>\n",
       "      <th>tipo_mudanca</th>\n",
       "      <th>window_size</th>\n",
       "      <th>AVG_roc_scores</th>\n",
       "      <th>AVG_avg_precisions</th>\n",
       "      <th>flag_binary_support</th>\n",
       "      <th>flag_frequency_support</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>kmeans__k=2</td>\n",
       "      <td>binary</td>\n",
       "      <td>False</td>\n",
       "      <td>cb</td>\n",
       "      <td>75</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>48</th>\n",
       "      <td>kmeans__k=2</td>\n",
       "      <td>binary</td>\n",
       "      <td>False</td>\n",
       "      <td>pm</td>\n",
       "      <td>75</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>54</th>\n",
       "      <td>kmeans__k=2</td>\n",
       "      <td>binary</td>\n",
       "      <td>False</td>\n",
       "      <td>re</td>\n",
       "      <td>75</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>60</th>\n",
       "      <td>kmeans__k=2</td>\n",
       "      <td>binary</td>\n",
       "      <td>False</td>\n",
       "      <td>rp</td>\n",
       "      <td>75</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>18</th>\n",
       "      <td>kmeans__k=2</td>\n",
       "      <td>binary</td>\n",
       "      <td>False</td>\n",
       "      <td>cm</td>\n",
       "      <td>75</td>\n",
       "      <td>0.990741</td>\n",
       "      <td>0.833333</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>211</th>\n",
       "      <td>kmeans__k=2</td>\n",
       "      <td>transitions</td>\n",
       "      <td>False</td>\n",
       "      <td>sw</td>\n",
       "      <td>100</td>\n",
       "      <td>0.939216</td>\n",
       "      <td>0.964583</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>229</th>\n",
       "      <td>kmeans__k=3</td>\n",
       "      <td>binary</td>\n",
       "      <td>False</td>\n",
       "      <td>cf</td>\n",
       "      <td>100</td>\n",
       "      <td>0.830688</td>\n",
       "      <td>0.696036</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>312</th>\n",
       "      <td>kmeans__k=3</td>\n",
       "      <td>frequency</td>\n",
       "      <td>False</td>\n",
       "      <td>cp</td>\n",
       "      <td>75</td>\n",
       "      <td>0.825132</td>\n",
       "      <td>0.723438</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>189</th>\n",
       "      <td>kmeans__k=2</td>\n",
       "      <td>transitions</td>\n",
       "      <td>False</td>\n",
       "      <td>pl</td>\n",
       "      <td>150</td>\n",
       "      <td>0.812203</td>\n",
       "      <td>0.882045</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>368</th>\n",
       "      <td>kmeans__k=3</td>\n",
       "      <td>transitions</td>\n",
       "      <td>False</td>\n",
       "      <td>cd</td>\n",
       "      <td>125</td>\n",
       "      <td>0.747318</td>\n",
       "      <td>0.761222</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>110</th>\n",
       "      <td>kmeans__k=2</td>\n",
       "      <td>frequency</td>\n",
       "      <td>False</td>\n",
       "      <td>lp</td>\n",
       "      <td>125</td>\n",
       "      <td>0.632576</td>\n",
       "      <td>0.710100</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>390</th>\n",
       "      <td>kmeans__k=3</td>\n",
       "      <td>transitions</td>\n",
       "      <td>False</td>\n",
       "      <td>fr</td>\n",
       "      <td>75</td>\n",
       "      <td>0.551163</td>\n",
       "      <td>0.633086</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "           model representation  sliding_window tipo_mudanca  window_size  \\\n",
       "0    kmeans__k=2         binary           False           cb           75   \n",
       "48   kmeans__k=2         binary           False           pm           75   \n",
       "54   kmeans__k=2         binary           False           re           75   \n",
       "60   kmeans__k=2         binary           False           rp           75   \n",
       "18   kmeans__k=2         binary           False           cm           75   \n",
       "211  kmeans__k=2    transitions           False           sw          100   \n",
       "229  kmeans__k=3         binary           False           cf          100   \n",
       "312  kmeans__k=3      frequency           False           cp           75   \n",
       "189  kmeans__k=2    transitions           False           pl          150   \n",
       "368  kmeans__k=3    transitions           False           cd          125   \n",
       "110  kmeans__k=2      frequency           False           lp          125   \n",
       "390  kmeans__k=3    transitions           False           fr           75   \n",
       "\n",
       "     AVG_roc_scores  AVG_avg_precisions  flag_binary_support  \\\n",
       "0          1.000000            1.000000                 True   \n",
       "48         1.000000            1.000000                 True   \n",
       "54         1.000000            1.000000                 True   \n",
       "60         1.000000            1.000000                 True   \n",
       "18         0.990741            0.833333                 True   \n",
       "211        0.939216            0.964583                False   \n",
       "229        0.830688            0.696036                 True   \n",
       "312        0.825132            0.723438                False   \n",
       "189        0.812203            0.882045                False   \n",
       "368        0.747318            0.761222                False   \n",
       "110        0.632576            0.710100                False   \n",
       "390        0.551163            0.633086                 True   \n",
       "\n",
       "     flag_frequency_support  \n",
       "0                      True  \n",
       "48                     True  \n",
       "54                     True  \n",
       "60                     True  \n",
       "18                     True  \n",
       "211                   False  \n",
       "229                    True  \n",
       "312                   False  \n",
       "189                   False  \n",
       "368                   False  \n",
       "110                    True  \n",
       "390                    True  "
      ]
     },
     "execution_count": 290,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "resp_df_grouped_2 = resp__2.groupby([\n",
    "    'model', 'representation', 'sliding_window', 'tipo_mudanca', 'window_size',\n",
    "]).agg({\n",
    "    'AVG_roc_scores': 'mean',\n",
    "    'AVG_avg_precisions': 'mean',\n",
    "    'flag_binary_support': 'max',\n",
    "    'flag_frequency_support': 'max',\n",
    "}).reset_index()\n",
    "\n",
    "resp_df_grouped_2.sort_values([\"AVG_roc_scores\", \"window_size\", 'representation', \"model\"],ascending=[False, True, True, True]).drop_duplicates(\"tipo_mudanca\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 291,
   "metadata": {},
   "outputs": [],
   "source": [
    "resp__2.to_excel(\"results_localization_experiments_TodaJanelaAnterior.xlsx\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 292,
   "metadata": {},
   "outputs": [],
   "source": [
    "resp__2.to_pickle(\"results_localization_experiments_TodaJanelaAnterior.pickle\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 295,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>model</th>\n",
       "      <th>representation</th>\n",
       "      <th>sliding_window</th>\n",
       "      <th>window_size</th>\n",
       "      <th>AVG_roc_scores</th>\n",
       "      <th>AVG_avg_precisions</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>26</th>\n",
       "      <td>kmeans__k=3</td>\n",
       "      <td>frequency</td>\n",
       "      <td>False</td>\n",
       "      <td>125</td>\n",
       "      <td>0.842959</td>\n",
       "      <td>0.811704</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>kmeans__k=2</td>\n",
       "      <td>frequency</td>\n",
       "      <td>False</td>\n",
       "      <td>100</td>\n",
       "      <td>0.840444</td>\n",
       "      <td>0.816752</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>kmeans__k=2</td>\n",
       "      <td>frequency</td>\n",
       "      <td>False</td>\n",
       "      <td>125</td>\n",
       "      <td>0.840178</td>\n",
       "      <td>0.812711</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>kmeans__k=2</td>\n",
       "      <td>frequency</td>\n",
       "      <td>False</td>\n",
       "      <td>250</td>\n",
       "      <td>0.838714</td>\n",
       "      <td>0.811968</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>kmeans__k=2</td>\n",
       "      <td>frequency</td>\n",
       "      <td>False</td>\n",
       "      <td>200</td>\n",
       "      <td>0.836858</td>\n",
       "      <td>0.807092</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>kmeans__k=2</td>\n",
       "      <td>frequency</td>\n",
       "      <td>False</td>\n",
       "      <td>75</td>\n",
       "      <td>0.836276</td>\n",
       "      <td>0.805166</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25</th>\n",
       "      <td>kmeans__k=3</td>\n",
       "      <td>frequency</td>\n",
       "      <td>False</td>\n",
       "      <td>100</td>\n",
       "      <td>0.834754</td>\n",
       "      <td>0.764308</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>kmeans__k=2</td>\n",
       "      <td>frequency</td>\n",
       "      <td>False</td>\n",
       "      <td>150</td>\n",
       "      <td>0.833989</td>\n",
       "      <td>0.803653</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>27</th>\n",
       "      <td>kmeans__k=3</td>\n",
       "      <td>frequency</td>\n",
       "      <td>False</td>\n",
       "      <td>150</td>\n",
       "      <td>0.829467</td>\n",
       "      <td>0.766574</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>29</th>\n",
       "      <td>kmeans__k=3</td>\n",
       "      <td>frequency</td>\n",
       "      <td>False</td>\n",
       "      <td>250</td>\n",
       "      <td>0.828848</td>\n",
       "      <td>0.797713</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>20</th>\n",
       "      <td>kmeans__k=3</td>\n",
       "      <td>binary</td>\n",
       "      <td>False</td>\n",
       "      <td>125</td>\n",
       "      <td>0.826988</td>\n",
       "      <td>0.793105</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>24</th>\n",
       "      <td>kmeans__k=3</td>\n",
       "      <td>frequency</td>\n",
       "      <td>False</td>\n",
       "      <td>75</td>\n",
       "      <td>0.826047</td>\n",
       "      <td>0.735524</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>28</th>\n",
       "      <td>kmeans__k=3</td>\n",
       "      <td>frequency</td>\n",
       "      <td>False</td>\n",
       "      <td>200</td>\n",
       "      <td>0.823455</td>\n",
       "      <td>0.754452</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>19</th>\n",
       "      <td>kmeans__k=3</td>\n",
       "      <td>binary</td>\n",
       "      <td>False</td>\n",
       "      <td>100</td>\n",
       "      <td>0.817970</td>\n",
       "      <td>0.759571</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>kmeans__k=2</td>\n",
       "      <td>binary</td>\n",
       "      <td>False</td>\n",
       "      <td>125</td>\n",
       "      <td>0.815156</td>\n",
       "      <td>0.783074</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>18</th>\n",
       "      <td>kmeans__k=3</td>\n",
       "      <td>binary</td>\n",
       "      <td>False</td>\n",
       "      <td>75</td>\n",
       "      <td>0.814936</td>\n",
       "      <td>0.745686</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>kmeans__k=2</td>\n",
       "      <td>binary</td>\n",
       "      <td>False</td>\n",
       "      <td>250</td>\n",
       "      <td>0.814050</td>\n",
       "      <td>0.782364</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>kmeans__k=2</td>\n",
       "      <td>binary</td>\n",
       "      <td>False</td>\n",
       "      <td>100</td>\n",
       "      <td>0.813731</td>\n",
       "      <td>0.781629</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>kmeans__k=2</td>\n",
       "      <td>binary</td>\n",
       "      <td>False</td>\n",
       "      <td>75</td>\n",
       "      <td>0.812781</td>\n",
       "      <td>0.780973</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>kmeans__k=2</td>\n",
       "      <td>binary</td>\n",
       "      <td>False</td>\n",
       "      <td>150</td>\n",
       "      <td>0.812466</td>\n",
       "      <td>0.780306</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>kmeans__k=2</td>\n",
       "      <td>binary</td>\n",
       "      <td>False</td>\n",
       "      <td>200</td>\n",
       "      <td>0.811037</td>\n",
       "      <td>0.779341</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>21</th>\n",
       "      <td>kmeans__k=3</td>\n",
       "      <td>binary</td>\n",
       "      <td>False</td>\n",
       "      <td>150</td>\n",
       "      <td>0.809618</td>\n",
       "      <td>0.743788</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>22</th>\n",
       "      <td>kmeans__k=3</td>\n",
       "      <td>binary</td>\n",
       "      <td>False</td>\n",
       "      <td>200</td>\n",
       "      <td>0.807068</td>\n",
       "      <td>0.756152</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>23</th>\n",
       "      <td>kmeans__k=3</td>\n",
       "      <td>binary</td>\n",
       "      <td>False</td>\n",
       "      <td>250</td>\n",
       "      <td>0.799466</td>\n",
       "      <td>0.764577</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>17</th>\n",
       "      <td>kmeans__k=2</td>\n",
       "      <td>transitions</td>\n",
       "      <td>False</td>\n",
       "      <td>250</td>\n",
       "      <td>0.799279</td>\n",
       "      <td>0.728370</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13</th>\n",
       "      <td>kmeans__k=2</td>\n",
       "      <td>transitions</td>\n",
       "      <td>False</td>\n",
       "      <td>100</td>\n",
       "      <td>0.796632</td>\n",
       "      <td>0.729674</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14</th>\n",
       "      <td>kmeans__k=2</td>\n",
       "      <td>transitions</td>\n",
       "      <td>False</td>\n",
       "      <td>125</td>\n",
       "      <td>0.796209</td>\n",
       "      <td>0.726379</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>16</th>\n",
       "      <td>kmeans__k=2</td>\n",
       "      <td>transitions</td>\n",
       "      <td>False</td>\n",
       "      <td>200</td>\n",
       "      <td>0.785875</td>\n",
       "      <td>0.720044</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>15</th>\n",
       "      <td>kmeans__k=2</td>\n",
       "      <td>transitions</td>\n",
       "      <td>False</td>\n",
       "      <td>150</td>\n",
       "      <td>0.785841</td>\n",
       "      <td>0.721777</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td>kmeans__k=2</td>\n",
       "      <td>transitions</td>\n",
       "      <td>False</td>\n",
       "      <td>75</td>\n",
       "      <td>0.784835</td>\n",
       "      <td>0.721751</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>31</th>\n",
       "      <td>kmeans__k=3</td>\n",
       "      <td>transitions</td>\n",
       "      <td>False</td>\n",
       "      <td>100</td>\n",
       "      <td>0.691051</td>\n",
       "      <td>0.609154</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>35</th>\n",
       "      <td>kmeans__k=3</td>\n",
       "      <td>transitions</td>\n",
       "      <td>False</td>\n",
       "      <td>250</td>\n",
       "      <td>0.686450</td>\n",
       "      <td>0.604740</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>32</th>\n",
       "      <td>kmeans__k=3</td>\n",
       "      <td>transitions</td>\n",
       "      <td>False</td>\n",
       "      <td>125</td>\n",
       "      <td>0.685563</td>\n",
       "      <td>0.595787</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>30</th>\n",
       "      <td>kmeans__k=3</td>\n",
       "      <td>transitions</td>\n",
       "      <td>False</td>\n",
       "      <td>75</td>\n",
       "      <td>0.678876</td>\n",
       "      <td>0.585356</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>33</th>\n",
       "      <td>kmeans__k=3</td>\n",
       "      <td>transitions</td>\n",
       "      <td>False</td>\n",
       "      <td>150</td>\n",
       "      <td>0.678443</td>\n",
       "      <td>0.598271</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>34</th>\n",
       "      <td>kmeans__k=3</td>\n",
       "      <td>transitions</td>\n",
       "      <td>False</td>\n",
       "      <td>200</td>\n",
       "      <td>0.670724</td>\n",
       "      <td>0.592456</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "          model representation  sliding_window  window_size  AVG_roc_scores  \\\n",
       "26  kmeans__k=3      frequency           False          125        0.842959   \n",
       "7   kmeans__k=2      frequency           False          100        0.840444   \n",
       "8   kmeans__k=2      frequency           False          125        0.840178   \n",
       "11  kmeans__k=2      frequency           False          250        0.838714   \n",
       "10  kmeans__k=2      frequency           False          200        0.836858   \n",
       "6   kmeans__k=2      frequency           False           75        0.836276   \n",
       "25  kmeans__k=3      frequency           False          100        0.834754   \n",
       "9   kmeans__k=2      frequency           False          150        0.833989   \n",
       "27  kmeans__k=3      frequency           False          150        0.829467   \n",
       "29  kmeans__k=3      frequency           False          250        0.828848   \n",
       "20  kmeans__k=3         binary           False          125        0.826988   \n",
       "24  kmeans__k=3      frequency           False           75        0.826047   \n",
       "28  kmeans__k=3      frequency           False          200        0.823455   \n",
       "19  kmeans__k=3         binary           False          100        0.817970   \n",
       "2   kmeans__k=2         binary           False          125        0.815156   \n",
       "18  kmeans__k=3         binary           False           75        0.814936   \n",
       "5   kmeans__k=2         binary           False          250        0.814050   \n",
       "1   kmeans__k=2         binary           False          100        0.813731   \n",
       "0   kmeans__k=2         binary           False           75        0.812781   \n",
       "3   kmeans__k=2         binary           False          150        0.812466   \n",
       "4   kmeans__k=2         binary           False          200        0.811037   \n",
       "21  kmeans__k=3         binary           False          150        0.809618   \n",
       "22  kmeans__k=3         binary           False          200        0.807068   \n",
       "23  kmeans__k=3         binary           False          250        0.799466   \n",
       "17  kmeans__k=2    transitions           False          250        0.799279   \n",
       "13  kmeans__k=2    transitions           False          100        0.796632   \n",
       "14  kmeans__k=2    transitions           False          125        0.796209   \n",
       "16  kmeans__k=2    transitions           False          200        0.785875   \n",
       "15  kmeans__k=2    transitions           False          150        0.785841   \n",
       "12  kmeans__k=2    transitions           False           75        0.784835   \n",
       "31  kmeans__k=3    transitions           False          100        0.691051   \n",
       "35  kmeans__k=3    transitions           False          250        0.686450   \n",
       "32  kmeans__k=3    transitions           False          125        0.685563   \n",
       "30  kmeans__k=3    transitions           False           75        0.678876   \n",
       "33  kmeans__k=3    transitions           False          150        0.678443   \n",
       "34  kmeans__k=3    transitions           False          200        0.670724   \n",
       "\n",
       "    AVG_avg_precisions  \n",
       "26            0.811704  \n",
       "7             0.816752  \n",
       "8             0.812711  \n",
       "11            0.811968  \n",
       "10            0.807092  \n",
       "6             0.805166  \n",
       "25            0.764308  \n",
       "9             0.803653  \n",
       "27            0.766574  \n",
       "29            0.797713  \n",
       "20            0.793105  \n",
       "24            0.735524  \n",
       "28            0.754452  \n",
       "19            0.759571  \n",
       "2             0.783074  \n",
       "18            0.745686  \n",
       "5             0.782364  \n",
       "1             0.781629  \n",
       "0             0.780973  \n",
       "3             0.780306  \n",
       "4             0.779341  \n",
       "21            0.743788  \n",
       "22            0.756152  \n",
       "23            0.764577  \n",
       "17            0.728370  \n",
       "13            0.729674  \n",
       "14            0.726379  \n",
       "16            0.720044  \n",
       "15            0.721777  \n",
       "12            0.721751  \n",
       "31            0.609154  \n",
       "35            0.604740  \n",
       "32            0.595787  \n",
       "30            0.585356  \n",
       "33            0.598271  \n",
       "34            0.592456  "
      ]
     },
     "execution_count": 295,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "resp_df_grouped_2[resp_df_grouped_2.flag_frequency_support].groupby(['model', 'representation', 'sliding_window', 'window_size']).agg({\n",
    "    'AVG_roc_scores': 'mean',\n",
    "    'AVG_avg_precisions': 'mean',\n",
    "}).reset_index().sort_values([\"AVG_roc_scores\", \"AVG_avg_precisions\"], ascending=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 298,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>model</th>\n",
       "      <th>representation</th>\n",
       "      <th>sliding_window</th>\n",
       "      <th>tipo_mudanca</th>\n",
       "      <th>window_size</th>\n",
       "      <th>AVG_roc_scores</th>\n",
       "      <th>AVG_avg_precisions</th>\n",
       "      <th>flag_binary_support</th>\n",
       "      <th>flag_frequency_support</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>290</th>\n",
       "      <td>kmeans__k=3</td>\n",
       "      <td>frequency</td>\n",
       "      <td>False</td>\n",
       "      <td>cb</td>\n",
       "      <td>125</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>296</th>\n",
       "      <td>kmeans__k=3</td>\n",
       "      <td>frequency</td>\n",
       "      <td>False</td>\n",
       "      <td>cd</td>\n",
       "      <td>125</td>\n",
       "      <td>0.430515</td>\n",
       "      <td>0.403129</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>302</th>\n",
       "      <td>kmeans__k=3</td>\n",
       "      <td>frequency</td>\n",
       "      <td>False</td>\n",
       "      <td>cf</td>\n",
       "      <td>125</td>\n",
       "      <td>0.804233</td>\n",
       "      <td>0.689454</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>308</th>\n",
       "      <td>kmeans__k=3</td>\n",
       "      <td>frequency</td>\n",
       "      <td>False</td>\n",
       "      <td>cm</td>\n",
       "      <td>125</td>\n",
       "      <td>0.982510</td>\n",
       "      <td>0.820988</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>314</th>\n",
       "      <td>kmeans__k=3</td>\n",
       "      <td>frequency</td>\n",
       "      <td>False</td>\n",
       "      <td>cp</td>\n",
       "      <td>125</td>\n",
       "      <td>0.787831</td>\n",
       "      <td>0.696037</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>320</th>\n",
       "      <td>kmeans__k=3</td>\n",
       "      <td>frequency</td>\n",
       "      <td>False</td>\n",
       "      <td>fr</td>\n",
       "      <td>125</td>\n",
       "      <td>0.359568</td>\n",
       "      <td>0.375722</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>326</th>\n",
       "      <td>kmeans__k=3</td>\n",
       "      <td>frequency</td>\n",
       "      <td>False</td>\n",
       "      <td>lp</td>\n",
       "      <td>125</td>\n",
       "      <td>0.622054</td>\n",
       "      <td>0.718583</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>332</th>\n",
       "      <td>kmeans__k=3</td>\n",
       "      <td>frequency</td>\n",
       "      <td>False</td>\n",
       "      <td>pl</td>\n",
       "      <td>125</td>\n",
       "      <td>0.380071</td>\n",
       "      <td>0.394774</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>338</th>\n",
       "      <td>kmeans__k=3</td>\n",
       "      <td>frequency</td>\n",
       "      <td>False</td>\n",
       "      <td>pm</td>\n",
       "      <td>125</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>344</th>\n",
       "      <td>kmeans__k=3</td>\n",
       "      <td>frequency</td>\n",
       "      <td>False</td>\n",
       "      <td>re</td>\n",
       "      <td>125</td>\n",
       "      <td>0.975309</td>\n",
       "      <td>0.888889</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>350</th>\n",
       "      <td>kmeans__k=3</td>\n",
       "      <td>frequency</td>\n",
       "      <td>False</td>\n",
       "      <td>rp</td>\n",
       "      <td>125</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>356</th>\n",
       "      <td>kmeans__k=3</td>\n",
       "      <td>frequency</td>\n",
       "      <td>False</td>\n",
       "      <td>sw</td>\n",
       "      <td>125</td>\n",
       "      <td>0.246528</td>\n",
       "      <td>0.163395</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "           model representation  sliding_window tipo_mudanca  window_size  \\\n",
       "290  kmeans__k=3      frequency           False           cb          125   \n",
       "296  kmeans__k=3      frequency           False           cd          125   \n",
       "302  kmeans__k=3      frequency           False           cf          125   \n",
       "308  kmeans__k=3      frequency           False           cm          125   \n",
       "314  kmeans__k=3      frequency           False           cp          125   \n",
       "320  kmeans__k=3      frequency           False           fr          125   \n",
       "326  kmeans__k=3      frequency           False           lp          125   \n",
       "332  kmeans__k=3      frequency           False           pl          125   \n",
       "338  kmeans__k=3      frequency           False           pm          125   \n",
       "344  kmeans__k=3      frequency           False           re          125   \n",
       "350  kmeans__k=3      frequency           False           rp          125   \n",
       "356  kmeans__k=3      frequency           False           sw          125   \n",
       "\n",
       "     AVG_roc_scores  AVG_avg_precisions  flag_binary_support  \\\n",
       "290        1.000000            1.000000                 True   \n",
       "296        0.430515            0.403129                False   \n",
       "302        0.804233            0.689454                 True   \n",
       "308        0.982510            0.820988                 True   \n",
       "314        0.787831            0.696037                False   \n",
       "320        0.359568            0.375722                 True   \n",
       "326        0.622054            0.718583                False   \n",
       "332        0.380071            0.394774                False   \n",
       "338        1.000000            1.000000                 True   \n",
       "344        0.975309            0.888889                 True   \n",
       "350        1.000000            1.000000                 True   \n",
       "356        0.246528            0.163395                False   \n",
       "\n",
       "     flag_frequency_support  \n",
       "290                    True  \n",
       "296                   False  \n",
       "302                    True  \n",
       "308                    True  \n",
       "314                   False  \n",
       "320                    True  \n",
       "326                    True  \n",
       "332                   False  \n",
       "338                    True  \n",
       "344                    True  \n",
       "350                    True  \n",
       "356                   False  "
      ]
     },
     "execution_count": 298,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "resp_df_grouped_2[\n",
    "    (resp_df_grouped_2.model == 'kmeans__k=3') &\n",
    "    (resp_df_grouped_2.representation == \"frequency\") &\n",
    "    (resp_df_grouped_2.window_size == 125)\n",
    "]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}