stack_12_1-18,20.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/javascript": [
       "IPython.notebook.set_autosave_interval(0)"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Autosave disabled\n"
     ]
    }
   ],
   "source": [
    "%autosave 0"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "import os\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "from sklearn.linear_model import Lasso, LassoCV, LinearRegression\n",
    "from sklearn.model_selection import KFold\n",
    "from utilities import cal_score"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "stack_idx = '12'\n",
    "models = '1-18,20'\n",
    "use_test_kfold = set([2, 7, 8, 12, 13])\n",
    "\n",
    "is_per_area = False"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Read CV predictions and test"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "def parse_models(exp):\n",
    "    exp_split = exp.split(',')\n",
    "    idx_models = []\n",
    "    for e in exp_split:\n",
    "        if '-' in e:\n",
    "            n0, n1 = e.split('-')\n",
    "            idx_models.extend(list(range(int(n0), int(n1)+1, 1)))\n",
    "        else:\n",
    "            idx_models.append(int(e))\n",
    "    return idx_models"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 20]\n"
     ]
    }
   ],
   "source": [
    "idx_models = parse_models(models)\n",
    "print(idx_models)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "files_in_output = [f for f in os.listdir('output/') if os.path.isfile('output/'+f)]\n",
    "files_cv = {idx: [f for f in files_in_output \n",
    "                  if f.startswith('model-%02d-' % idx) and f.endswith('cv.csv')][0] \n",
    "            for idx in idx_models}\n",
    "files_test_one = {idx: [f for f in files_in_output \n",
    "                        if f.startswith('model-%02d-' % idx) and f.endswith('test-one.csv')][0]\n",
    "                  for idx in idx_models}\n",
    "files_test_kf = {idx: [f for f in files_in_output \n",
    "                       if f.startswith('model-%02d-' % idx) and f.endswith('test-kfold.csv')][0]\n",
    "                 for idx in idx_models}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      " 1 model-01-lgb-cv.csv\n",
      " 1 model-01-lgb-test-kfold.csv\n",
      " 1 model-01-lgb-test-one.csv\n",
      " 2 model-02-keras-search-cv.csv\n",
      " 2 model-02-keras-search-test-kfold.csv\n",
      " 2 model-02-keras-search-test-one.csv\n",
      " 3 model-03-lgb-feats-selection-cv.csv\n",
      " 3 model-03-lgb-feats-selection-test-kfold.csv\n",
      " 3 model-03-lgb-feats-selection-test-one.csv\n",
      " 4 model-04-lgb-PCA-cv.csv\n",
      " 4 model-04-lgb-PCA-test-kfold.csv\n",
      " 4 model-04-lgb-PCA-test-one.csv\n",
      " 5 model-05-lgb-wo-per-area-cv.csv\n",
      " 5 model-05-lgb-wo-per-area-test-kfold.csv\n",
      " 5 model-05-lgb-wo-per-area-test-one.csv\n",
      " 6 model-06-lgb-lr0.001-cv.csv\n",
      " 6 model-06-lgb-lr0.001-test-kfold.csv\n",
      " 6 model-06-lgb-lr0.001-test-one.csv\n",
      " 7 model-07-keras-embedding-cv.csv\n",
      " 7 model-07-keras-embedding-test-kfold.csv\n",
      " 7 model-07-keras-embedding-test-one.csv\n",
      " 8 model-08-keras-search-long-cv.csv\n",
      " 8 model-08-keras-search-long-test-kfold.csv\n",
      " 8 model-08-keras-search-long-test-one.csv\n",
      " 9 model-09-lgb-feats-selection-75-cv.csv\n",
      " 9 model-09-lgb-feats-selection-75-test-kfold.csv\n",
      " 9 model-09-lgb-feats-selection-75-test-one.csv\n",
      "10 model-10-lgb-feats-selection-75-lr-0.001-cv.csv\n",
      "10 model-10-lgb-feats-selection-75-lr-0.001-test-kfold.csv\n",
      "10 model-10-lgb-feats-selection-75-lr-0.001-test-one.csv\n",
      "11 model-11-rf-cv.csv\n",
      "11 model-11-rf-test-kfold.csv\n",
      "11 model-11-rf-test-one.csv\n",
      "12 model-12-predict-keras-search-prelu-cv.csv\n",
      "12 model-12-predict-keras-search-prelu-test-kfold.csv\n",
      "12 model-12-predict-keras-search-prelu-test-one.csv\n",
      "13 model-13-predict-keras-he_uni-cv.csv\n",
      "13 model-13-predict-keras-he_uni-test-kfold.csv\n",
      "13 model-13-predict-keras-he_uni-test-one.csv\n",
      "14 model-14-lgb-feats-selection-75-lr-0.001-rand-cv.csv\n",
      "14 model-14-lgb-feats-selection-75-lr-0.001-rand-test-kfold.csv\n",
      "14 model-14-lgb-feats-selection-75-lr-0.001-rand-test-one.csv\n",
      "15 model-15-lgb-feats-selection-75-lr-0.001-rand323-cv.csv\n",
      "15 model-15-lgb-feats-selection-75-lr-0.001-rand323-test-kfold.csv\n",
      "15 model-15-lgb-feats-selection-75-lr-0.001-rand323-test-one.csv\n",
      "16 model-16-lgb-feats-selection-68-lr-0.001-mix5-cv.csv\n",
      "16 model-16-lgb-feats-selection-68-lr-0.001-mix5-test-kfold.csv\n",
      "16 model-16-lgb-feats-selection-68-lr-0.001-mix5-test-one.csv\n",
      "17 model-17-lgb-feats-selection-70-lr-0.001-mix5-cv.csv\n",
      "17 model-17-lgb-feats-selection-70-lr-0.001-mix5-test-kfold.csv\n",
      "17 model-17-lgb-feats-selection-70-lr-0.001-mix5-test-one.csv\n",
      "18 model-18-lgb-feats-selection-70-lr-0.001-p5-cv.csv\n",
      "18 model-18-lgb-feats-selection-70-lr-0.001-p5-test-kfold.csv\n",
      "18 model-18-lgb-feats-selection-70-lr-0.001-p5-test-one.csv\n",
      "20 model-20-lgb-lr-0.0008-mix5-cv.csv\n",
      "20 model-20-lgb-lr-0.0008-mix5-test-kfold.csv\n",
      "20 model-20-lgb-lr-0.0008-mix5-test-one.csv\n"
     ]
    }
   ],
   "source": [
    "for k in files_cv: \n",
    "    print('%2d'%k, files_cv[k])\n",
    "    print('%2d'%k, files_test_kf[k])\n",
    "    print('%2d'%k, files_test_one[k])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Load area"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "df_train = pd.read_csv('dataset/train.csv')\n",
    "df_test = pd.read_csv('dataset/test.csv')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "cv = df_train[['building_id', 'building_area', 'total_price']]\n",
    "test = df_test[['building_id', 'building_area']]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CV predictions:\n",
      "No. 0 file: model-01-lgb-cv.csv\n",
      "No. 1 file: model-02-keras-search-cv.csv\n",
      "No. 2 file: model-03-lgb-feats-selection-cv.csv\n",
      "No. 3 file: model-04-lgb-PCA-cv.csv\n",
      "No. 4 file: model-05-lgb-wo-per-area-cv.csv\n",
      "No. 5 file: model-06-lgb-lr0.001-cv.csv\n",
      "No. 6 file: model-07-keras-embedding-cv.csv\n",
      "No. 7 file: model-08-keras-search-long-cv.csv\n",
      "No. 8 file: model-09-lgb-feats-selection-75-cv.csv\n",
      "No. 9 file: model-10-lgb-feats-selection-75-lr-0.001-cv.csv\n",
      "No. 10 file: model-11-rf-cv.csv\n",
      "No. 11 file: model-12-predict-keras-search-prelu-cv.csv\n",
      "No. 12 file: model-13-predict-keras-he_uni-cv.csv\n",
      "No. 13 file: model-14-lgb-feats-selection-75-lr-0.001-rand-cv.csv\n",
      "No. 14 file: model-15-lgb-feats-selection-75-lr-0.001-rand323-cv.csv\n",
      "No. 15 file: model-16-lgb-feats-selection-68-lr-0.001-mix5-cv.csv\n",
      "No. 16 file: model-17-lgb-feats-selection-70-lr-0.001-mix5-cv.csv\n",
      "No. 17 file: model-18-lgb-feats-selection-70-lr-0.001-p5-cv.csv\n",
      "No. 18 file: model-20-lgb-lr-0.0008-mix5-cv.csv\n"
     ]
    }
   ],
   "source": [
    "print('CV predictions:')\n",
    "for i, idx_model in enumerate(idx_models):\n",
    "    f = files_cv[idx_model]\n",
    "    print('No. {} file: {}'.format(i, f))\n",
    "    df = pd.read_csv('output/'+f)\n",
    "    \n",
    "    cv = pd.merge(cv, df[['building_id', 'total_price_predict']], on='building_id')\n",
    "    \n",
    "    cv = cv.rename(columns = {'total_price_predict': 'pred_{}'.format(idx_model)})\n",
    "    cv[f'log_pred_{idx_model}'] = np.log1p(cv[f'pred_{idx_model}'])\n",
    "    cv[f'log_parea_pred_{idx_model}'] = np.log1p( cv[f'pred_{idx_model}'] / cv['building_area'] )"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Test predictions:\n",
      "No. 0 file: model-01-lgb-test-one.csv\n",
      "No. 1 file: model-02-keras-search-test-kfold.csv\n",
      "No. 2 file: model-03-lgb-feats-selection-test-one.csv\n",
      "No. 3 file: model-04-lgb-PCA-test-one.csv\n",
      "No. 4 file: model-05-lgb-wo-per-area-test-one.csv\n",
      "No. 5 file: model-06-lgb-lr0.001-test-one.csv\n",
      "No. 6 file: model-07-keras-embedding-test-kfold.csv\n",
      "No. 7 file: model-08-keras-search-long-test-kfold.csv\n",
      "No. 8 file: model-09-lgb-feats-selection-75-test-one.csv\n",
      "No. 9 file: model-10-lgb-feats-selection-75-lr-0.001-test-one.csv\n",
      "No. 10 file: model-11-rf-test-one.csv\n",
      "No. 11 file: model-12-predict-keras-search-prelu-test-kfold.csv\n",
      "No. 12 file: model-13-predict-keras-he_uni-test-kfold.csv\n",
      "No. 13 file: model-14-lgb-feats-selection-75-lr-0.001-rand-test-one.csv\n",
      "No. 14 file: model-15-lgb-feats-selection-75-lr-0.001-rand323-test-one.csv\n",
      "No. 15 file: model-16-lgb-feats-selection-68-lr-0.001-mix5-test-one.csv\n",
      "No. 16 file: model-17-lgb-feats-selection-70-lr-0.001-mix5-test-one.csv\n",
      "No. 17 file: model-18-lgb-feats-selection-70-lr-0.001-p5-test-one.csv\n",
      "No. 18 file: model-20-lgb-lr-0.0008-mix5-test-one.csv\n"
     ]
    }
   ],
   "source": [
    "cv['log_total_price'] = np.log1p(cv['total_price'])\n",
    "cv['log_parea_total_price'] = np.log1p( cv['total_price'] / cv['building_area'] )\n",
    "\n",
    "print('Test predictions:')\n",
    "for i, idx in enumerate(idx_models):\n",
    "    f = files_test_kf[idx] if idx in use_test_kfold else files_test_one[idx]\n",
    "    print('No. {} file: {}'.format(i, f))\n",
    "    df = pd.read_csv('output/'+f)\n",
    "\n",
    "    test = pd.merge(test, df[['building_id','total_price']], on='building_id')\n",
    "        \n",
    "    test = test.rename(columns = {'total_price': 'pred_{}'.format(idx)})\n",
    "    test[f'log_pred_{idx}'] = np.log1p(test[f'pred_{idx}'])\n",
    "    test[f'log_parea_pred_{idx}'] = np.log1p( test[f'pred_{idx}'] / test['building_area'] )"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "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>building_id</th>\n",
       "      <th>building_area</th>\n",
       "      <th>total_price</th>\n",
       "      <th>pred_1</th>\n",
       "      <th>log_pred_1</th>\n",
       "      <th>log_parea_pred_1</th>\n",
       "      <th>pred_2</th>\n",
       "      <th>log_pred_2</th>\n",
       "      <th>log_parea_pred_2</th>\n",
       "      <th>pred_3</th>\n",
       "      <th>...</th>\n",
       "      <th>log_pred_17</th>\n",
       "      <th>log_parea_pred_17</th>\n",
       "      <th>pred_18</th>\n",
       "      <th>log_pred_18</th>\n",
       "      <th>log_parea_pred_18</th>\n",
       "      <th>pred_20</th>\n",
       "      <th>log_pred_20</th>\n",
       "      <th>log_parea_pred_20</th>\n",
       "      <th>log_total_price</th>\n",
       "      <th>log_parea_total_price</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>e3mMIMR3JJqCaXz1</td>\n",
       "      <td>3.418175</td>\n",
       "      <td>6.476038e+05</td>\n",
       "      <td>6.331552e+05</td>\n",
       "      <td>13.358472</td>\n",
       "      <td>12.129369</td>\n",
       "      <td>717209.94</td>\n",
       "      <td>13.483125</td>\n",
       "      <td>12.254022</td>\n",
       "      <td>6.656891e+05</td>\n",
       "      <td>...</td>\n",
       "      <td>13.392766</td>\n",
       "      <td>12.163663</td>\n",
       "      <td>6.631879e+05</td>\n",
       "      <td>13.404815</td>\n",
       "      <td>12.175712</td>\n",
       "      <td>6.385893e+05</td>\n",
       "      <td>13.367018</td>\n",
       "      <td>12.137915</td>\n",
       "      <td>13.381036</td>\n",
       "      <td>12.151933</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>LgwzgklNvy4QCtq5</td>\n",
       "      <td>4.041309</td>\n",
       "      <td>3.321452e+06</td>\n",
       "      <td>3.064324e+06</td>\n",
       "      <td>14.935338</td>\n",
       "      <td>13.538770</td>\n",
       "      <td>2899842.20</td>\n",
       "      <td>14.880167</td>\n",
       "      <td>13.483600</td>\n",
       "      <td>3.079196e+06</td>\n",
       "      <td>...</td>\n",
       "      <td>14.956603</td>\n",
       "      <td>13.560036</td>\n",
       "      <td>3.092403e+06</td>\n",
       "      <td>14.944459</td>\n",
       "      <td>13.547892</td>\n",
       "      <td>3.171011e+06</td>\n",
       "      <td>14.969561</td>\n",
       "      <td>13.572994</td>\n",
       "      <td>15.015913</td>\n",
       "      <td>13.619345</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>ucIR2NLLsC3T650L</td>\n",
       "      <td>5.584279</td>\n",
       "      <td>9.570885e+06</td>\n",
       "      <td>9.827776e+06</td>\n",
       "      <td>16.100723</td>\n",
       "      <td>14.380769</td>\n",
       "      <td>9766813.00</td>\n",
       "      <td>16.094501</td>\n",
       "      <td>14.374546</td>\n",
       "      <td>9.814852e+06</td>\n",
       "      <td>...</td>\n",
       "      <td>16.096597</td>\n",
       "      <td>14.376642</td>\n",
       "      <td>9.773138e+06</td>\n",
       "      <td>16.095148</td>\n",
       "      <td>14.375193</td>\n",
       "      <td>9.802951e+06</td>\n",
       "      <td>16.098194</td>\n",
       "      <td>14.378239</td>\n",
       "      <td>16.074236</td>\n",
       "      <td>14.354282</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>jre1pJhcQj91Kdky</td>\n",
       "      <td>13.563031</td>\n",
       "      <td>1.421501e+07</td>\n",
       "      <td>1.255350e+07</td>\n",
       "      <td>16.345510</td>\n",
       "      <td>13.738164</td>\n",
       "      <td>12699800.00</td>\n",
       "      <td>16.357097</td>\n",
       "      <td>13.749750</td>\n",
       "      <td>1.255981e+07</td>\n",
       "      <td>...</td>\n",
       "      <td>16.357815</td>\n",
       "      <td>13.750469</td>\n",
       "      <td>1.266263e+07</td>\n",
       "      <td>16.354166</td>\n",
       "      <td>13.746819</td>\n",
       "      <td>1.266383e+07</td>\n",
       "      <td>16.354260</td>\n",
       "      <td>13.746914</td>\n",
       "      <td>16.469809</td>\n",
       "      <td>13.862462</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>rQpYpY9nRG7X5mmr</td>\n",
       "      <td>4.688108</td>\n",
       "      <td>7.627120e+05</td>\n",
       "      <td>1.215194e+06</td>\n",
       "      <td>14.010415</td>\n",
       "      <td>12.465389</td>\n",
       "      <td>2012610.50</td>\n",
       "      <td>14.514944</td>\n",
       "      <td>12.969916</td>\n",
       "      <td>1.128419e+06</td>\n",
       "      <td>...</td>\n",
       "      <td>13.919882</td>\n",
       "      <td>12.374856</td>\n",
       "      <td>1.128016e+06</td>\n",
       "      <td>13.935971</td>\n",
       "      <td>12.390946</td>\n",
       "      <td>1.124958e+06</td>\n",
       "      <td>13.933257</td>\n",
       "      <td>12.388231</td>\n",
       "      <td>13.544637</td>\n",
       "      <td>11.999613</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 62 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "        building_id  building_area   total_price        pred_1  log_pred_1  \\\n",
       "0  e3mMIMR3JJqCaXz1       3.418175  6.476038e+05  6.331552e+05   13.358472   \n",
       "1  LgwzgklNvy4QCtq5       4.041309  3.321452e+06  3.064324e+06   14.935338   \n",
       "2  ucIR2NLLsC3T650L       5.584279  9.570885e+06  9.827776e+06   16.100723   \n",
       "3  jre1pJhcQj91Kdky      13.563031  1.421501e+07  1.255350e+07   16.345510   \n",
       "4  rQpYpY9nRG7X5mmr       4.688108  7.627120e+05  1.215194e+06   14.010415   \n",
       "\n",
       "   log_parea_pred_1       pred_2  log_pred_2  log_parea_pred_2        pred_3  \\\n",
       "0         12.129369    717209.94   13.483125         12.254022  6.656891e+05   \n",
       "1         13.538770   2899842.20   14.880167         13.483600  3.079196e+06   \n",
       "2         14.380769   9766813.00   16.094501         14.374546  9.814852e+06   \n",
       "3         13.738164  12699800.00   16.357097         13.749750  1.255981e+07   \n",
       "4         12.465389   2012610.50   14.514944         12.969916  1.128419e+06   \n",
       "\n",
       "   ...  log_pred_17  log_parea_pred_17       pred_18  log_pred_18  \\\n",
       "0  ...    13.392766          12.163663  6.631879e+05    13.404815   \n",
       "1  ...    14.956603          13.560036  3.092403e+06    14.944459   \n",
       "2  ...    16.096597          14.376642  9.773138e+06    16.095148   \n",
       "3  ...    16.357815          13.750469  1.266263e+07    16.354166   \n",
       "4  ...    13.919882          12.374856  1.128016e+06    13.935971   \n",
       "\n",
       "   log_parea_pred_18       pred_20  log_pred_20  log_parea_pred_20  \\\n",
       "0          12.175712  6.385893e+05    13.367018          12.137915   \n",
       "1          13.547892  3.171011e+06    14.969561          13.572994   \n",
       "2          14.375193  9.802951e+06    16.098194          14.378239   \n",
       "3          13.746819  1.266383e+07    16.354260          13.746914   \n",
       "4          12.390946  1.124958e+06    13.933257          12.388231   \n",
       "\n",
       "   log_total_price  log_parea_total_price  \n",
       "0        13.381036              12.151933  \n",
       "1        15.015913              13.619345  \n",
       "2        16.074236              14.354282  \n",
       "3        16.469809              13.862462  \n",
       "4        13.544637              11.999613  \n",
       "\n",
       "[5 rows x 62 columns]"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "cv.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "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>building_id</th>\n",
       "      <th>building_area</th>\n",
       "      <th>pred_1</th>\n",
       "      <th>log_pred_1</th>\n",
       "      <th>log_parea_pred_1</th>\n",
       "      <th>pred_2</th>\n",
       "      <th>log_pred_2</th>\n",
       "      <th>log_parea_pred_2</th>\n",
       "      <th>pred_3</th>\n",
       "      <th>log_pred_3</th>\n",
       "      <th>...</th>\n",
       "      <th>log_parea_pred_16</th>\n",
       "      <th>pred_17</th>\n",
       "      <th>log_pred_17</th>\n",
       "      <th>log_parea_pred_17</th>\n",
       "      <th>pred_18</th>\n",
       "      <th>log_pred_18</th>\n",
       "      <th>log_parea_pred_18</th>\n",
       "      <th>pred_20</th>\n",
       "      <th>log_pred_20</th>\n",
       "      <th>log_parea_pred_20</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>X5gsdTWGS3W7JJQB</td>\n",
       "      <td>3.418175</td>\n",
       "      <td>1.526912e+07</td>\n",
       "      <td>16.541343</td>\n",
       "      <td>15.312236</td>\n",
       "      <td>12470072.0</td>\n",
       "      <td>16.338842</td>\n",
       "      <td>15.109735</td>\n",
       "      <td>1.531685e+07</td>\n",
       "      <td>16.544464</td>\n",
       "      <td>...</td>\n",
       "      <td>15.217310</td>\n",
       "      <td>1.449905e+07</td>\n",
       "      <td>16.489594</td>\n",
       "      <td>15.260487</td>\n",
       "      <td>1.396801e+07</td>\n",
       "      <td>16.452280</td>\n",
       "      <td>15.223173</td>\n",
       "      <td>1.478805e+07</td>\n",
       "      <td>16.509330</td>\n",
       "      <td>15.280223</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>BTshNOJyKHnT2YIT</td>\n",
       "      <td>7.726227</td>\n",
       "      <td>3.924241e+06</td>\n",
       "      <td>15.182684</td>\n",
       "      <td>13.138065</td>\n",
       "      <td>3916552.2</td>\n",
       "      <td>15.180723</td>\n",
       "      <td>13.136104</td>\n",
       "      <td>3.977095e+06</td>\n",
       "      <td>15.196062</td>\n",
       "      <td>...</td>\n",
       "      <td>13.136639</td>\n",
       "      <td>3.917960e+06</td>\n",
       "      <td>15.181082</td>\n",
       "      <td>13.136463</td>\n",
       "      <td>3.908782e+06</td>\n",
       "      <td>15.178737</td>\n",
       "      <td>13.134118</td>\n",
       "      <td>3.918591e+06</td>\n",
       "      <td>15.181243</td>\n",
       "      <td>13.136624</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>dhdymr0lV8N5kZOT</td>\n",
       "      <td>12.170581</td>\n",
       "      <td>1.096127e+07</td>\n",
       "      <td>16.209879</td>\n",
       "      <td>13.710858</td>\n",
       "      <td>11912735.0</td>\n",
       "      <td>16.293119</td>\n",
       "      <td>13.794098</td>\n",
       "      <td>1.084967e+07</td>\n",
       "      <td>16.199646</td>\n",
       "      <td>...</td>\n",
       "      <td>13.690042</td>\n",
       "      <td>1.063343e+07</td>\n",
       "      <td>16.179513</td>\n",
       "      <td>13.680493</td>\n",
       "      <td>1.073427e+07</td>\n",
       "      <td>16.188952</td>\n",
       "      <td>13.689931</td>\n",
       "      <td>1.069408e+07</td>\n",
       "      <td>16.185201</td>\n",
       "      <td>13.686181</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>VEwyGGMcD56w5BOc</td>\n",
       "      <td>2.252256</td>\n",
       "      <td>6.155550e+06</td>\n",
       "      <td>15.632865</td>\n",
       "      <td>14.820933</td>\n",
       "      <td>5940670.0</td>\n",
       "      <td>15.597333</td>\n",
       "      <td>14.785400</td>\n",
       "      <td>6.015238e+06</td>\n",
       "      <td>15.609807</td>\n",
       "      <td>...</td>\n",
       "      <td>14.808737</td>\n",
       "      <td>6.119193e+06</td>\n",
       "      <td>15.626941</td>\n",
       "      <td>14.815009</td>\n",
       "      <td>6.084894e+06</td>\n",
       "      <td>15.621320</td>\n",
       "      <td>14.809388</td>\n",
       "      <td>6.033528e+06</td>\n",
       "      <td>15.612843</td>\n",
       "      <td>14.800910</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>wmUeMoJZfsqaSX9b</td>\n",
       "      <td>5.813985</td>\n",
       "      <td>1.062995e+06</td>\n",
       "      <td>13.876602</td>\n",
       "      <td>12.116340</td>\n",
       "      <td>1088488.1</td>\n",
       "      <td>13.900301</td>\n",
       "      <td>12.140039</td>\n",
       "      <td>1.027248e+06</td>\n",
       "      <td>13.842395</td>\n",
       "      <td>...</td>\n",
       "      <td>12.125206</td>\n",
       "      <td>1.084776e+06</td>\n",
       "      <td>13.896885</td>\n",
       "      <td>12.136623</td>\n",
       "      <td>1.087941e+06</td>\n",
       "      <td>13.899798</td>\n",
       "      <td>12.139536</td>\n",
       "      <td>1.089268e+06</td>\n",
       "      <td>13.901017</td>\n",
       "      <td>12.140755</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 59 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "        building_id  building_area        pred_1  log_pred_1  \\\n",
       "0  X5gsdTWGS3W7JJQB       3.418175  1.526912e+07   16.541343   \n",
       "1  BTshNOJyKHnT2YIT       7.726227  3.924241e+06   15.182684   \n",
       "2  dhdymr0lV8N5kZOT      12.170581  1.096127e+07   16.209879   \n",
       "3  VEwyGGMcD56w5BOc       2.252256  6.155550e+06   15.632865   \n",
       "4  wmUeMoJZfsqaSX9b       5.813985  1.062995e+06   13.876602   \n",
       "\n",
       "   log_parea_pred_1      pred_2  log_pred_2  log_parea_pred_2        pred_3  \\\n",
       "0         15.312236  12470072.0   16.338842         15.109735  1.531685e+07   \n",
       "1         13.138065   3916552.2   15.180723         13.136104  3.977095e+06   \n",
       "2         13.710858  11912735.0   16.293119         13.794098  1.084967e+07   \n",
       "3         14.820933   5940670.0   15.597333         14.785400  6.015238e+06   \n",
       "4         12.116340   1088488.1   13.900301         12.140039  1.027248e+06   \n",
       "\n",
       "   log_pred_3  ...  log_parea_pred_16       pred_17  log_pred_17  \\\n",
       "0   16.544464  ...          15.217310  1.449905e+07    16.489594   \n",
       "1   15.196062  ...          13.136639  3.917960e+06    15.181082   \n",
       "2   16.199646  ...          13.690042  1.063343e+07    16.179513   \n",
       "3   15.609807  ...          14.808737  6.119193e+06    15.626941   \n",
       "4   13.842395  ...          12.125206  1.084776e+06    13.896885   \n",
       "\n",
       "   log_parea_pred_17       pred_18  log_pred_18  log_parea_pred_18  \\\n",
       "0          15.260487  1.396801e+07    16.452280          15.223173   \n",
       "1          13.136463  3.908782e+06    15.178737          13.134118   \n",
       "2          13.680493  1.073427e+07    16.188952          13.689931   \n",
       "3          14.815009  6.084894e+06    15.621320          14.809388   \n",
       "4          12.136623  1.087941e+06    13.899798          12.139536   \n",
       "\n",
       "        pred_20  log_pred_20  log_parea_pred_20  \n",
       "0  1.478805e+07    16.509330          15.280223  \n",
       "1  3.918591e+06    15.181243          13.136624  \n",
       "2  1.069408e+07    16.185201          13.686181  \n",
       "3  6.033528e+06    15.612843          14.800910  \n",
       "4  1.089268e+06    13.901017          12.140755  \n",
       "\n",
       "[5 rows x 59 columns]"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "test.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Check models scores"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      " 0 model-01 5870.873059\n",
      " 1 model-02 5400.852164\n",
      " 2 model-03 5877.873452\n",
      " 3 model-04 5713.867808\n",
      " 4 model-05 5724.869598\n",
      " 5 model-06 5886.873769\n",
      " 6 model-07 5171.836449\n",
      " 7 model-08 5514.858826\n",
      " 8 model-09 5872.873118\n",
      " 9 model-10 5897.873845\n",
      "10 model-11 5075.838018\n",
      "11 model-12 5486.856963\n",
      "12 model-13 5506.858055\n",
      "13 model-14 5908.873901\n",
      "14 model-15 5900.873836\n",
      "15 model-16 5907.874126\n",
      "16 model-17 5905.874165\n",
      "17 model-18 5908.874297\n",
      "18 model-20 5908.874040\n"
     ]
    }
   ],
   "source": [
    "for i, idx_model in enumerate(idx_models):\n",
    "    print('%2d'%i, 'model-%02d'%idx_model, '%.6f'%cal_score(cv['total_price'], cv[f'pred_{idx_model}']))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Make Xy"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "if is_per_area:\n",
    "    X = cv[['log_parea_pred_{}'.format(idx) for idx in idx_models]]\n",
    "else:\n",
    "    X = cv[['log_pred_{}'.format(idx) for idx in idx_models]]\n",
    "\n",
    "if is_per_area:\n",
    "    y = cv['log_parea_total_price']\n",
    "else:\n",
    "    y = cv['log_total_price']"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### start regression"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "#reg = LassoCV(alphas=[0]+list(np.logspace(-4, 3, 7)), max_iter=100000, tol=1e-6, n_jobs=-1)\n",
    "#reg.fit(X, y)\n",
    "\n",
    "#print(reg.alpha_)\n",
    "#print(reg.mse_path_)\n",
    "#print(reg.coef_, reg.intercept_)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [],
   "source": [
    "#for a in [0]+list(np.logspace(-4, 3, 7)):\n",
    "#    reg_single = Lasso(alpha=a, max_iter=100000, tol=1e-6)\n",
    "#    reg_single.fit(X, y)\n",
    "#    print(reg_single.coef_, reg.intercept_)\n",
    "#    print(reg_single.score(X,y))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "==== Fold 1 ====\n",
      "alpha, score: 0 5989.874905595002\n",
      "alpha, score: 1e-05 5998.874964313814\n",
      "alpha, score: 2e-05 5998.874977805055\n",
      "alpha, score: 5e-05 5992.87497440904\n",
      "alpha, score: 8e-05 5991.8749776784325\n",
      "alpha, score: 0.0001 5989.874981089196\n",
      "alpha, score: 0.0002 5996.874990653028\n",
      "alpha, score: 0.0005 5980.874932804801\n",
      "alpha, score: 0.0008 5980.874747996822\n",
      "alpha, score: 0.001 5980.874717211465\n",
      "alpha, score: 0.002 5981.87473064469\n",
      "alpha, score: 0.005 5980.874719092851\n",
      "alpha, score: 0.008 5970.874617582694\n",
      "alpha, score: 0.01 5968.874498339265\n",
      "alpha, score: 0.02 5944.87342744609\n",
      "==== Fold 2 ====\n",
      "alpha, score: 0 5929.875627116668\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/usr/local/lib/python3.7/site-packages/sklearn/linear_model/coordinate_descent.py:492: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations. Fitting data with very small alpha may cause precision problems.\n",
      "  ConvergenceWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "alpha, score: 1e-05 5932.875722123361\n",
      "alpha, score: 2e-05 5933.8757518219\n",
      "alpha, score: 5e-05 5933.875790780346\n",
      "alpha, score: 8e-05 5928.875797545809\n",
      "alpha, score: 0.0001 5926.875806084041\n",
      "alpha, score: 0.0002 5928.875842710134\n",
      "alpha, score: 0.0005 5933.875875987377\n",
      "alpha, score: 0.0008 5933.875798489939\n",
      "alpha, score: 0.001 5938.875724116016\n",
      "alpha, score: 0.002 5938.875745799369\n",
      "alpha, score: 0.005 5936.875755940565\n",
      "alpha, score: 0.008 5928.87568048495\n",
      "alpha, score: 0.01 5933.8755852935465\n",
      "alpha, score: 0.02 5907.874635339946\n",
      "==== Fold 3 ====\n",
      "alpha, score: 0 5888.874585186534\n",
      "alpha, score: 1e-05 5888.874633523749\n",
      "alpha, score: 2e-05 5899.8746490783315\n",
      "alpha, score: 5e-05 5901.874649588399\n",
      "alpha, score: 8e-05 5900.874659404939\n",
      "alpha, score: 0.0001 5899.874665380619\n",
      "alpha, score: 0.0002 5896.874687147786\n",
      "alpha, score: 0.0005 5910.874675487289\n",
      "alpha, score: 0.0008 5890.874530382214\n",
      "alpha, score: 0.001 5884.874454465015\n",
      "alpha, score: 0.002 5883.874463646796\n",
      "alpha, score: 0.005 5878.874436862272\n",
      "alpha, score: 0.008 5874.874322559933\n",
      "alpha, score: 0.01 5873.874193508168\n",
      "alpha, score: 0.02 5844.873063577523\n",
      "[2e-05, 5944.208459568428, [5998.874977805055, 5933.8757518219, 5899.8746490783315]]\n",
      "[5e-05, 5942.875138259263, [5992.87497440904, 5933.875790780346, 5901.874649588399]]\n",
      "[0.0005, 5941.875161426488, [5980.874932804801, 5933.875875987377, 5910.874675487289]]\n",
      "[0.0002, 5940.875173503649, [5996.874990653028, 5928.875842710134, 5896.874687147786]]\n",
      "[8e-05, 5940.541811543061, [5991.8749776784325, 5928.875797545809, 5900.874659404939]]\n",
      "[1e-05, 5940.208439986975, [5998.874964313814, 5932.875722123361, 5888.874633523749]]\n",
      "[0.0001, 5938.875150851286, [5989.874981089196, 5926.875806084041, 5899.874665380619]]\n",
      "[0, 5936.2083726327355, [5989.874905595002, 5929.875627116668, 5888.874585186534]]\n",
      "[0.0008, 5935.208358956325, [5980.874747996822, 5933.875798489939, 5890.874530382214]]\n",
      "[0.002, 5934.874980030286, [5981.87473064469, 5938.875745799369, 5883.874463646796]]\n",
      "[0.001, 5934.8749652641645, [5980.874717211465, 5938.875724116016, 5884.874454465015]]\n",
      "[0.005, 5932.208303965229, [5980.874719092851, 5936.875755940565, 5878.874436862272]]\n",
      "[0.01, 5925.541425713659, [5968.874498339265, 5933.8755852935465, 5873.874193508168]]\n",
      "[0.008, 5924.874873542525, [5970.874617582694, 5928.87568048495, 5874.874322559933]]\n",
      "[0.02, 5899.207042121186, [5944.87342744609, 5907.874635339946, 5844.873063577523]]\n"
     ]
    }
   ],
   "source": [
    "alphas = [0, 0.00001, 0.00002, 0.00005, 0.00008, 0.0001, 0.0002, 0.0005, 0.0008, 0.001, 0.002, 0.005, 0.008, 0.01, 0.02]\n",
    "gsearch = {}\n",
    "\n",
    "folds = KFold(n_splits=3, shuffle=True, random_state=1208)\n",
    "for i_fold, (itrain, ival) in enumerate(folds.split(X)): # kfold\n",
    "    print('==== Fold', i_fold+1, '====')\n",
    "    \n",
    "    # split train, val\n",
    "    X_train = X.iloc[itrain]\n",
    "    X_val = X.iloc[ival]\n",
    "    y_train = y.iloc[itrain]\n",
    "    y_val = y.iloc[ival]\n",
    "    \n",
    "    # random sample - grid search\n",
    "    for a in alphas:\n",
    "        if a == 0:\n",
    "            reg_single = LinearRegression()\n",
    "        else:\n",
    "            reg_single = Lasso(alpha=a, max_iter=100000, tol=1e-6)\n",
    "        reg_single.fit(X_train, y_train)\n",
    "        \n",
    "        y_pred = reg_single.predict(X_val)\n",
    "        if is_per_area:\n",
    "            y_pred_final = np.expm1(y_pred) * cv.iloc[ival]['building_area']\n",
    "            y_true_final = np.expm1(y_val) * cv.iloc[ival]['building_area']\n",
    "        else:\n",
    "            y_pred_final = np.expm1(y_pred)\n",
    "            y_true_final = np.expm1(y_val)\n",
    "        score = cal_score(y_true_final, y_pred_final)\n",
    "\n",
    "        print('alpha, score:', a, score)\n",
    "        gsearch[a] = gsearch.get(a, []) + [score]\n",
    "\n",
    "results = [[key, np.mean(value), value] for key, value, in gsearch.items()]\n",
    "results.sort(key= lambda x: x[1], reverse=True)\n",
    "for item in results:\n",
    "    print(item)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2e-05\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "Lasso(alpha=2e-05, copy_X=True, fit_intercept=True, max_iter=1000000,\n",
       "   normalize=False, positive=False, precompute=False, random_state=None,\n",
       "   selection='cyclic', tol=1e-06, warm_start=False)"
      ]
     },
     "execution_count": 19,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "alpha_set = results[0][0]\n",
    "print(alpha_set)\n",
    "if alpha_set == 0:\n",
    "    reg = LinearRegression()\n",
    "else:\n",
    "    reg = Lasso(alpha=alpha_set, max_iter=1000000, tol=1e-6)\n",
    "reg.fit(X, y)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[-0.13610486  0.04578194  0.10943289  0.01835466  0.21948601 -0.\n",
      "  0.10049487  0.02569241 -0.          0.         -0.11752065  0.02757014\n",
      "  0.01264898  0.09400721  0.          0.13073791  0.          0.03321083\n",
      "  0.4391602 ] -0.04400248840710752\n"
     ]
    }
   ],
   "source": [
    "print(reg.coef_, reg.intercept_)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Calculate cv score"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [],
   "source": [
    "cv_pred_final = np.zeros(X.shape[0])\n",
    "for i, col in enumerate(X):\n",
    "    cv_pred_final = cv_pred_final + X[col] * reg.coef_[i]\n",
    "cv_pred_final = cv_pred_final + reg.intercept_\n",
    "\n",
    "if is_per_area:\n",
    "    cv_pred_final = np.expm1(cv_pred_final) * cv['building_area']\n",
    "    cv_true_final = np.expm1(y) * cv['building_area']\n",
    "else:\n",
    "    cv_pred_final = np.expm1(cv_pred_final)\n",
    "    cv_true_final = np.expm1(y)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "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",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>6.476038e+05</td>\n",
       "      <td>6.207041e+05</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>3.321452e+06</td>\n",
       "      <td>3.143776e+06</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>9.570885e+06</td>\n",
       "      <td>9.804127e+06</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>1.421501e+07</td>\n",
       "      <td>1.281052e+07</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>7.627120e+05</td>\n",
       "      <td>1.263111e+06</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "              a             b\n",
       "0  6.476038e+05  6.207041e+05\n",
       "1  3.321452e+06  3.143776e+06\n",
       "2  9.570885e+06  9.804127e+06\n",
       "3  1.421501e+07  1.281052e+07\n",
       "4  7.627120e+05  1.263111e+06"
      ]
     },
     "execution_count": 22,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pd.DataFrame({'a':cv_true_final,'b':cv_pred_final}).head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "5943.875351218801"
      ]
     },
     "execution_count": 23,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "cal_score(cv_true_final, cv_pred_final)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "33.5320091454936\n"
     ]
    }
   ],
   "source": [
    "y_pred_cv = reg.predict(X)\n",
    "y_pred_final = np.expm1(y_pred_cv) * cv['building_area']\n",
    "y_true_final = cv['total_price']\n",
    "print(cal_score(y_true_final, y_pred_final))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Compute submission"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [],
   "source": [
    "if is_per_area:\n",
    "    col_prefix = 'log_parea_pred'\n",
    "else:\n",
    "    col_prefix = 'log_pred'\n",
    "\n",
    "test_pred_final = pd.DataFrame({'building_id': test['building_id'], 'total_price': np.zeros(test.shape[0])})\n",
    "\n",
    "for i, idx in enumerate(idx_models):\n",
    "    test_pred_final['total_price'] = test_pred_final['total_price'] + test[f'{col_prefix}_{idx}'] * reg.coef_[i]\n",
    "test_pred_final['total_price'] = test_pred_final['total_price'] + reg.intercept_\n",
    "\n",
    "if is_per_area:\n",
    "    test_pred_final['total_price'] = np.expm1(test_pred_final['total_price']) * test['building_area'] \n",
    "else:\n",
    "    test_pred_final['total_price'] = np.expm1(test_pred_final['total_price'])\n",
    "    \n",
    "test_pred_final['total_price'] = np.clip(test_pred_final['total_price'], 0, None)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [],
   "source": [
    "import matplotlib.pyplot as plt"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/usr/local/lib/python3.7/site-packages/matplotlib/axes/_axes.py:6521: MatplotlibDeprecationWarning: \n",
      "The 'normed' kwarg was deprecated in Matplotlib 2.1 and will be removed in 3.1. Use 'density' instead.\n",
      "  alternative=\"'density'\", removal=\"3.1\")\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.hist(cv['log_parea_total_price'], bins=100, label='train true', normed=True)\n",
    "plt.hist(np.log1p(test_pred_final['total_price'] / test['building_area']), bins=100, label='test',\n",
    "         normed=True, alpha=0.7)\n",
    "plt.xlabel('log(price/area + 1)'); plt.ylabel('ratio')\n",
    "plt.legend(); plt.grid(); plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.hist(cv['log_total_price'], bins=100, label='train true', normed=True)\n",
    "plt.hist(np.log1p(test_pred_final['total_price']), bins=100, label='test', normed=True, alpha=0.7)\n",
    "plt.xlabel('log(price + 1)'); plt.ylabel('ratio')\n",
    "plt.legend(); plt.grid(); plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.hist(np.log1p(cv['building_area']), bins=100, label='train', normed=True)\n",
    "plt.hist(np.log1p(test['building_area']), bins=100, label='test', normed=True, alpha=0.7)\n",
    "plt.xlabel('log(building_area + 1)'); plt.ylabel('ratio'); plt.yscale('log')\n",
    "plt.legend(); plt.grid(); plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [],
   "source": [
    "if is_per_area:\n",
    "    test_pred_final.to_csv('output/stack_parea_{}_{}.csv'.format(stack_idx, models), index=False)\n",
    "else:\n",
    "    test_pred_final.to_csv('output/stack_{}_{}.csv'.format(stack_idx, models), index=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [],
   "source": [
    "#a= pd.read_csv('output/model-03-lgb-feats-selection-cv.csv')\n",
    "#b= pd.read_csv('output/model-03-lgb-feats-selection-test-one.csv')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [],
   "source": [
    "#a.rename(columns={'building_id':'id','total_price_predict':'target'}).to_csv('opt-pred3.csv',index=False)\n",
    "#b.rename(columns={'building_id':'id','total_price':'target'}).to_csv('opt-test3.csv',index=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "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": "3.7.3"
  },
  "toc": {
   "base_numbering": 1,
   "nav_menu": {},
   "number_sections": true,
   "sideBar": true,
   "skip_h1_title": true,
   "title_cell": "Table of Contents",
   "title_sidebar": "Contents",
   "toc_cell": false,
   "toc_position": {},
   "toc_section_display": true,
   "toc_window_display": true
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}