Predict Crops_v2.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "collapsed": true,
    "deletable": true,
    "editable": true
   },
   "outputs": [],
   "source": [
    "#from mpl_toolkits.basemap import Basemap\n",
    "\n",
    "import pandas as pd\n",
    "import numpy as np \n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "import psycopg2\n",
    "import seaborn as sns\n",
    "from sklearn.model_selection import train_test_split\n",
    "\n",
    "# from shapely.wkt import loads as wkt_loads\n",
    "from matplotlib.collections import LineCollection\n",
    "import matplotlib.cm as cm\n",
    "# import geopandas as gp\n",
    "\n",
    "# import h2o\n",
    "# from h2o.estimators.gbm import H2OGradientBoostingEstimator\n",
    "%matplotlib inline \n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {
    "collapsed": false,
    "deletable": true,
    "editable": true
   },
   "outputs": [],
   "source": [
    "try:\n",
    "    conn = psycopg2.connect(\"dbname='crops' user='postgres' host='localhost' port='5433'\")\n",
    "    cur = conn.cursor()\n",
    "except:\n",
    "    print \"Unable to connect to the database\""
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "deletable": true,
    "editable": true
   },
   "source": [
    "### EDA"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "collapsed": false,
    "deletable": true,
    "editable": true
   },
   "outputs": [],
   "source": [
    "def clean_values(x):\n",
    "    if type(x) != str:\n",
    "        return x\n",
    "    \n",
    "    x = x.lower()\n",
    "    \n",
    "    if \"aardappel\" in x:\n",
    "        return \"potato\" \n",
    "    \n",
    "    \n",
    "    elif any(s in x for s in [\"maïs\" ,\"mais\"]):\n",
    "        return \"corn\"\n",
    "\n",
    "    elif any(s in x for s in ['gerst','gierst','tarwe','rogge','haver',\\\n",
    "                              'triticale','graan','granen','spelt','boekwei']):\n",
    "        return \"grain\"    \n",
    "\n",
    "    elif any(s in x for s in [\"zaad\" ,\"zade\"]):\n",
    "        return \"seed\"    \n",
    "        \n",
    "    \n",
    "    elif \"gras\" in  x:\n",
    "        return \"grass\"\n",
    "    \n",
    "    \n",
    "    elif \"ui\" in x:\n",
    "        return \"onion\"\n",
    "    \n",
    "    \n",
    "\n",
    "    \n",
    "    elif any(s in x for s in ['bone','erwt','kapucijne']):\n",
    "        return \"beans\"\n",
    "   \n",
    "    \n",
    "    elif any(s in x for s in ['boom','bomen','note','fruit','vrucht',\\\n",
    "                              'appel','peren','kerst','kers','hout']):\n",
    "        return \"tree\"\n",
    "    \n",
    "    elif any(s in x for s in ['bloem','klaver','koolzaad','raapzaad','tulp',\\\n",
    "                              'luzerne','tagete','cichorei','lupinen','sierconiferen'\\\n",
    "                              ,'buxus','sierheester','ericaceae','azalea','snijgroen',\n",
    "                             'facelia','hennep','vlas','vezel','wikke','vaste planten']):\n",
    "        return \"flower\"\n",
    "    \n",
    "    elif any(s in x for s in ['groenten','peen','wortel','prei','witlof','kool',\\\n",
    "                              'asperges','rabarber','sla','knolselderij','biet',\\\n",
    "                             'pompoen','spinazie','broccoli',\\\n",
    "                             'bladrammenas','radijs','peulen','mosterd',\\\n",
    "                             'schorseneren','andijvie','knolvenkel','courgette','pastinaak',\n",
    "                             'valeriaan','knoflook','peterselie','paksoi',\\\n",
    "                              'bes','aardbei','frambo','selder','bram','braam',\\\n",
    "                             'komkommer','bladgewas']):\n",
    "        return \"veggies\"\n",
    "    \n",
    "\n",
    "\n",
    "    elif any(s in x for s in ['fauna','natuur','bos','dummy','onbeteelde','sloot'\\\n",
    "                              ,'heide','onbekend','gegeven','open grond']):\n",
    "        return \"waste\"\n",
    "    \n",
    "    else: return 'others'\n",
    "    \n",
    "    \n",
    "#recursively join data with previous years\n",
    "def get_backward_join_of_parcels(year,limit=2009,reference_year=False):\n",
    "    \n",
    "    #as id'+ str(year) + '\n",
    "    if reference_year:\n",
    "        q = 'select gid, cell_id, soil, next_year_parcel_fk as fk, gws_gewas as c'+ str(year) + ' from \"' + str(year) + '\"'\n",
    "    else: \n",
    "        q = 'select gid , next_year_parcel_fk  as fk, gws_gewas as c'+ str(year) + ' from \"' + str(year) + '\"'\n",
    "        \n",
    "        \n",
    "        \n",
    "    df =  pd.read_sql_query(q,conn)\n",
    "    \n",
    "    #clean crop values without messing other columns\n",
    "    if reference_year : \n",
    "        tempcols = df[['soil',\"cell_id\"]]\n",
    "        df = df.drop(['soil',\"cell_id\"],axis=1).applymap(clean_values)\n",
    "        df.insert(0,\"cell_id\",tempcols.cell_id )\n",
    "        df.insert(0,\"soil\",tempcols.soil )\n",
    "\n",
    "    else:\n",
    "        df = df.applymap(clean_values)\n",
    "    #end cleaning\n",
    "    \n",
    "    \n",
    "    \n",
    "        \n",
    "    if year > limit:\n",
    "        nextdf = get_backward_join_of_parcels(year-1,limit)\n",
    "        nextdf = nextdf.drop('gid',axis=1)\n",
    "        merged = df.merge(nextdf,how=\"left\",left_on = 'gid', right_on = 'fk')\n",
    "        merged.drop('fk_y',axis=1, inplace=True)\n",
    "        merged.rename(columns={'fk_x': 'fk'}, inplace=True)\n",
    "        \n",
    "            \n",
    "        \n",
    "        return merged\n",
    "    \n",
    "    else: #edge case, end recursion\n",
    "        return df\n",
    "\n",
    "# df15 = get_backward_join_of_parcels(2015)\n",
    "\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "collapsed": false,
    "deletable": true,
    "editable": true
   },
   "outputs": [],
   "source": [
    "df10 = get_backward_join_of_parcels(2010,reference_year=True)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {
    "collapsed": false,
    "deletable": true,
    "editable": true
   },
   "outputs": [],
   "source": [
    "all_years_df = pd.DataFrame()\n",
    "for year in range(2009,2017):\n",
    "    solodf = get_backward_join_of_parcels(year,year)\n",
    "    solodf = solodf.fillna('others')\n",
    "    all_years_df.insert(0,str(year),solodf['c'+str(year)].value_counts())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 107,
   "metadata": {
    "collapsed": false,
    "deletable": true,
    "editable": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "grass      495571.0\n",
       "corn       100981.0\n",
       "grain       44943.0\n",
       "potato      31760.0\n",
       "onion       31250.0\n",
       "waste       20144.0\n",
       "seed        20152.0\n",
       "tree        14609.0\n",
       "flower      12127.0\n",
       "others        505.0\n",
       "beans         823.0\n",
       "veggies         NaN\n",
       "Name: 2012, dtype: float64"
      ]
     },
     "execution_count": 107,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "all_years_df['2012']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 111,
   "metadata": {
    "collapsed": false,
    "deletable": true,
    "editable": true,
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.legend.Legend at 0x125576e90>"
      ]
     },
     "execution_count": 111,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAA74AAAHfCAYAAAB3SQTXAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\niIiIiKxpCr4iIiIiIiKypv3/iijs3Hki7XwAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x11c9bfc10>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig, ax = plt.subplots(figsize=(16,8))\n",
    "\n",
    "melteddf = pd.melt(all_years_df.reset_index(),id_vars=[\"index\"], \n",
    "                  var_name=\"Date\", value_name=\"Value\")\n",
    "\n",
    "melteddf = melteddf.sort_values(\"Date\", axis=0)\n",
    "melteddf = melteddf[melteddf[\"index\"]\\\n",
    "                    .apply((lambda x: False if any(s in x for s in [\"tree\",\"waste\",\"grass\",\"others\"]) else True))]\n",
    "\n",
    "for label, df in melteddf.groupby('index'):\n",
    "    df.plot(kind=\"line\",x=\"Date\",y=\"Value\", ax=ax, label=label)\n",
    "# bp = p_df.groupby('class').plot(kind='kde', ax=ax)\n",
    "\n",
    "plt.legend()    \n",
    "# # bp = melteddf.groupby('index').plot(kind='line',x=\"Date\", ax=ax,by='index')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 127,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>index</th>\n",
       "      <th>Date</th>\n",
       "      <th>Value</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>87</th>\n",
       "      <td>potato</td>\n",
       "      <td>2009</td>\n",
       "      <td>34078.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>75</th>\n",
       "      <td>potato</td>\n",
       "      <td>2010</td>\n",
       "      <td>34107.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>63</th>\n",
       "      <td>potato</td>\n",
       "      <td>2011</td>\n",
       "      <td>34407.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>51</th>\n",
       "      <td>potato</td>\n",
       "      <td>2012</td>\n",
       "      <td>31760.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>39</th>\n",
       "      <td>potato</td>\n",
       "      <td>2013</td>\n",
       "      <td>32978.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>27</th>\n",
       "      <td>potato</td>\n",
       "      <td>2014</td>\n",
       "      <td>33282.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>15</th>\n",
       "      <td>potato</td>\n",
       "      <td>2015</td>\n",
       "      <td>33437.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>potato</td>\n",
       "      <td>2016</td>\n",
       "      <td>33922.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "     index  Date    Value\n",
       "87  potato  2009  34078.0\n",
       "75  potato  2010  34107.0\n",
       "63  potato  2011  34407.0\n",
       "51  potato  2012  31760.0\n",
       "39  potato  2013  32978.0\n",
       "27  potato  2014  33282.0\n",
       "15  potato  2015  33437.0\n",
       "3   potato  2016  33922.0"
      ]
     },
     "execution_count": 127,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "melteddf[melteddf['index']==\"potato\"]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 113,
   "metadata": {
    "collapsed": false,
    "deletable": true,
    "editable": true
   },
   "outputs": [],
   "source": [
    "df16 = get_backward_join_of_parcels(2016,2009,True)\n",
    "\n",
    "# df14 = get_backward_join_of_parcels(2014,2009,True)\n",
    "\n",
    "    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 138,
   "metadata": {
    "collapsed": false,
    "deletable": true,
    "editable": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x11de303d0>"
      ]
     },
     "execution_count": 138,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYYAAAD3CAYAAAAZifM1AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAAFrBJREFUeJzt3XucXWV97/FPuEiwBE0hltKjHKT4AzQ9nHoBlKs1L0RF\nKJZWCiKCooittFBBS72WVl4IbakaAcHIsQaFohaRS0UToIBYgR65/fCUixQbbkXBKoGQnD+eNS/m\nmZ1kJrPX3jOZ9Xn/s2evvfbaz/q9YH7zrJX9fGetWrUKSZJGbDDVA5AkTS82BklSxcYgSarYGCRJ\nFRuDJKmy0VQPoA0rVjyz6rHHfjHVw5g25s59LtajZk1q1qNXF2syb96cWavbPiNmDBtttOFUD2Fa\nsR69rEnNevSyJs+aEY1BktQeG4MkqWJjkCRVbAySpIqNQZJUabUxRMT8iNiz+fneiJjd5vElSYPX\n9ozhLcBOLR9TkjREsya77HZEbAycB2wHbAh8FvgY8BRwGPBV4LvAts1bfhf4OfA5YHtKUzo5M5dE\nxK3AXcBy4NPA6cDTwGPAoZn5xDjDce1wSVp3q/2CWz/ffH438Ehmvi0i5gA3Ad8Ebs3MGyMC4NzM\nvDYiFgELgC2a9xwVEVsAVwMvBTYDPpGZN0fEacDFwKeANwNzgfEaAw8/PO4unTFv3hzrMYY1qVmP\nXl2sybx5c1a7vZ9LSTtSfrHT/EV/O2X2MNoPmsdlwHOB+cAbImIJ8I/ARk2DAMjm8a+AFwBXAb9H\nmTlIkoakn8ZwB7AHQDNjmA9cP+aYYy/x3Akszsy9gf2ACymXiwBWNo+HAosycx/gNuDoPsYoSVpH\n/TSGs4EtIuJaYAnl/sIPgPdFxD5reM9ZwA4RsRS4DrgvM1eO2ef7wBebfV4LnN/HGCVJ62jSN5+n\nmVVduza4Nl28Vjoea1KzHr26WJMZvbqqJKk9NgZJUsXGIEmq2BgkSRUbgySpYmOQJFVsDJKkio1B\nklSZEY1h/+O/MdVDkKQZY0Y0BklSe/pqDBFxRER8sq3BSJKmnjMGSVKln6CeEbtFxFXA5sBHKSlt\npwDPAP9OCfTZFPg88HxgS+CczFzY5DLcArysef/BwIOU9LfnNe/7QGYuaWGckqQJaKMx/DfwRmAe\n8D1KsM7umflQRHwCOIKyHPcFmXlxRGwNLAUWNu+/MTOPi4hTgEOAS4CtgNdRAnteMpFBrCmJqKus\nRy9rUrMevaxJ0UZjuDYzVwEPRcQvKRnPX22iPTcFrgQuBY6LiIOAx4GNR73/5ubxfmCrzLwtIj4D\nLG72O3Mig+jacrlr08Xlg8djTWrWo1cXazKIaM8RrwSIiK2A2cC9wAFNStspwHeBE4DrM/MwSmrb\n6DXAq0CIiJgPzMnMNwJvB/6+hTFKkiaojRnDphHxHWAz4F3AhsClEbEBZXZwOOWX/8KIOBR4FFgR\nEZus4Xg/Aj4SEYcDTwEfbmGMkqQJMsFtBurilHg81qRmPXp1sSYmuEmSJsTGIEmq2BgkSRUbgySp\nYmOQJFVsDJKkio1BklSxMUiSKjOiMex//Dc48pPfmephSNKMMCMagySpPdO6MUTEsqkegyR1zbRu\nDJKk4WtjddVKRLwEWEQJ7FlBWV31fcCelEZ0RmZe2CyvfSZlCe5HgSMp6W9nAy+lpL+taQVWSdKA\ntN4YgAWUxLY/BfYADgK2zczXRMRs4IaI+GfgHODIzLw9Io4CPgBcD8zOzF0j4kXA763LB5u+9Cxr\n0cua1KxHL2tSDKIxnAucCFwO/IyS6fzyJt8ZSirbNsCOwGebpLeNgbsoM4UbATLzxxFx/7p8cNeW\nzF2TLi4fPB5rUrMevbpYk0EmuI11AHBNZv4OJa3tHcB3m0S31wJfBe4GEji82f4BSvznncBuAE02\n9G8MYHySpLUYxIzhX4EvRcQKYCXlctChEXENJeXta5n5REQcA5wfERs27zsqM++KiN0j4nvAfcAj\nAxifJGktTHCbgbo4JR6PNalZj15drIkJbpKkCbExSJIqNgZJUsXGIEmq2BgkSRUbgySpYmOQJFVs\nDJKkyiC++Tx0+x//jbW+ft5Jrx3SSCRp/eeMQZJUmfLGEBGzI+KdUz0OSVIx5Y0B2AqwMUjSNNHX\nPYaIuAl4PfAYJYVtr8y8udl+BfAKYA5wR2a+IyJeA5xOSXd7DDgU+HNgp4j4MPB3lDyHLZqP+OPM\n/GE/Y5QkrZt+bz5/HdgX+A/gHmBBRCwH7gUey8wFEbEBcFtE/AZwIHAx8CngzcBc4BRgfmZ+PCJO\nBa7KzIURsT3wBWD3PsfYyVSmLp7zeKxJzXr0siZFv43hYspf/D9uHv+YcnlqMbBLRCym5DhvRklp\n+6tmv6uAB4DvUec6zwdeGxF/0Dyf2+f4gO4lu3Vx+eDxWJOa9ejVxZoMJMEtM28FtgVeBXyL0gAO\nAJYDL8zMQ4APAZsCsyiXjhZl5j7AbcDRlDCfkXHcCfxNk+r2+8A/9DM+SdK6a+Pm81Lg4cxc2fz8\nECW3+cURcQNwESXKc2vg+8AXI2IpJebz/Gb/5zSXkU4Bfr/Jh74cuLWF8UmS1oEJbjNQF6fE47Em\nNevRq4s1McFNkjQhNgZJUsXGIEmq2BgkSRUbgySpYmOQJFVsDJKkio1BklQxwU2SVHHGIEmqDL0x\nRMRJEfGqYX+uJGlihn4pKTM/OezPlCRN3FobQ0RcDPxdZi6NiFcCHwWWAdtTZhsnZ+aSiHgT8HHg\nZ5Rktv8LfAz4DCXFbRllee79m2NcQMlk+NxqjnUKZeXVDYDFmfm3bZ6wJGntxpsxnAO8nbKc9hGU\npbD/R2YeFRFbAFdHxG8BZwK7ZeaDETGSofBmYIvMfFVEzAN+NObY7wQeGX0s4KXA4cCewE+az+xb\nF1OZunjO47EmNevRy5oU4zWGK4DTIuJXgT0of8XvHhG7jHr/VsDjmflgs+2aZtuOwPUAmflwRNw5\n5tjzgT1GH6tpEG8F/ro5xmWTPrNROriUbufOeTzWpGY9enWxJpNKcGvCdy4EFlLyne+gXN7ZG9iv\nee0/gTnNrABg1+bxVmA3gIiYC7xkzOHvXM2xfg4cDBxCuZx0RERsM8FzlCS1YCL/Kuk84KDm8Sxg\nhyaB7TrgvqZ5vA/4VkR8G3gR8DRwKfBIRFwHnAv8otk+YnXHWg78F3AL8B3gSkqetCRpSFpJcIuI\nDwJnZObyiPgS5Rf6jcDOmXlBc4noNmCb5pd/20xwG6WLU+LxWJOa9ejVxZqsKcGtrX+u+gRwQ0T8\nArgX+Epz7FMj4jhgQ+DEATUFSVKLWmkMmflp4NNjNi8HDmjj+JKk4XFJDElSxcYgSarYGCRJFRuD\nJKliY5AkVWwMkqRKJxLcwBQ3SZooZwySpMpQGkNEzI6Idw7jsyRJ/RnWjGErSv6CJGmaG9Y9hj8H\ndoqIlcC3gc2Ao4DXAX8IrAIuyMwzI+KFwNnAbOBJ4OjMvH9I45SkzhtWYziFEsxzOTA3M98fETsB\nfwDsTmkM346IKygRoWdm5mUR8TvAJ4FD+x1A15KZuna+E2FNatajlzUppuJfJWXz+DJgG0r2M8Bc\n4DcpDeRDEXEiMAt4qo0P7dJyul1cPng81qRmPXp1sSaTSnBr0cpRn7WyeUxKRsM+TYrbIuCHlGS3\nE5tt7wYuGtIYJUkMb8bwEPAcYNORDZn5bxFxFXBtRGxCCfZ5ADgBWBgRs5v93z+kMUqSGFJjyMwn\ngZ1Xs/004LQxm+8G9h3GuCRJvWbEN58vOf2Azl0blKRB8ZvPkqSKjUGSVLExSJIqNgZJUsXGIEmq\n2BgkSRUbgySpYmOQJFVmxBfcJhLtOZZRn5K0en01hojYEPgW8CvApZn5162MSpI0Zfq9lPTrwJaU\n8J3H+h+OJGmq9Xsp6Wxge0qDWAYQEadTwncAvgx8CbgqM3eOiN2AS4F5zXvOBd4EfK45zgbAyZm5\nJCJuBe4ClmfmIX2OU5I0Qf02hvcCFwD/CRARbwK2BXZtjn0t8B3g0Say8/XA/cDLgVcAX6NkQT+S\nmUdFxBbA1cBLKfGfn8jMm/sc42rN9KSmmX5+k2FNatajlzUp2r75vCNwTWauAp6OiBuAnSgN4A3A\nq4FTgQXNz0cBHwb2iIhdRsbUNAh4Nu2tdTN5NdYuJlGNx5rUrEevLtZkWAlud9BcRoqIjSm//H8E\nfB34Q+Bx4DLgQGCTzFxGSWxb3CS27QdcyLP3K1YiSRqqVhtDZn4TuCcirgduAC7KzJsy8z+A2ZR7\nDY8BKyj3GgDOAnaIiKXAdcB9mWlDkKQpMmvVqlVTPYa+7X/8N9b5JGby9xi6OCUejzWpWY9eXazJ\nvHlzZq1u+4z4gpsJbpLUHpfEkCRVbAySpIqNQZJUsTFIkio2BklSxcYgSarYGCRJFRuDJKkyI77g\nNpkEtxEz+RvQkjQZzhgkSZWhNoaI2CoiPjvMz5QkrZuhXkpqltl+7zA/U5K0blppDE32wnnAdsCG\nwBnAMcAtwMuAzYGDgVnABZm5a0QsAP4SeBJ4FDgS2Bk4EXiKkgT3lcw8pY0xrslMTWyaqefVD2tS\nsx69rEnR1ozh3ZR4zrdFxBzgJmA5cGNmHhcRpwCHUGJAiYhZlLzo3TPzgYh4P3Ay8E1gG+C3gE2A\nnwADbQwzcVXWLi4fPB5rUrMevbpYk0EnuO1IyWomM58AbqfMHkbymu+nBPWM2BJ4PDMfaJ6P5DwD\n/DAzV2TmfwO/bGl8kqQJaqsx3AHsAdDMGOYD9wBrCtB5BNg8In69eb4XcFfz8/qfHCRJ67G2GsPZ\nwBYRcS2wBPgY8NCads7MVcC7gIsj4l+A1wGfaGkskqQ+zIhoT2BV164Nrk0Xr5WOx5rUrEevLtZk\nTdGefsFNklSxMUiSKjYGSVLFxiBJqtgYJEkVG4MkqWJjkCRVbAySpErnE9xWx1Q3SV3mjEGSVBlK\nY4iI943z+vyI2HMYY5Ekrd2wZgwnj/P6W4CdhjEQSdLaTfoeQ0QcARxASWfbEvg48Di9qWzHAr/a\nZD2fBHweeH7znnOAfwKOAJ6KiJuA5409Rmb+dLLjnIyZkOI0E86hbdakZj16WZOi35vPmwELgHnA\njcBKxqSyZeYJEfFHmfneiPhtSrTnxRGxNbA0MxdGxCJgGfB94O6xxwBO6HOc62R9X2Gxi6tEjsea\n1KxHry7WZFAJbkszc2VmPgj8HHhqDalsI5YBB0bElyi/8Dce8/rakt0kSUPQb2N4OUBE/BrwXOA5\na0hlG1nz+wTg+sw8DLhw1PaVzVjWluwmSRqCfi8lbRURV1HuCxwDrKCksq0EHqPcOwC4vZklnAss\njIhDKfcPVkTEJsAPgNMoEaHvWsMxJElDMOkEt+bm8w6ZeVKrI5ocE9xG6eK10vFYk5r16NXFmpjg\nJkmakElfSsrMRS2OQ5I0TThjkCRVbAySpIqNQZJUsTFIkio2BklSxcYgSaqY4LYOTHaT1AXOGCRJ\nlWnTGCLibyPiRVM9DknqumlzKSkzj5vqMUiSBtAYImJT4Hxga+B+YE/K0tkPA3MpMZ7nMCrFrQnr\nWQK8B3grsC3wAmAb4E8y84q2xylJWr1BzBiOBu7JzIMjYgfgNkpj+HJmfm11KW7AwjHHWJ6Z+0XE\nAuB4YFo0hvUp9m99GuuwWJOa9ehlTYpBNIYdgcsBMvPOiHi42Z7N4zLguIg4iJIRPTbFDeDm5vF+\nYPYAxjgp68uSvF1cPng81qRmPXp1sSaDivZcnVuB3QAiYjvK5SIoKW2w5hS30SYXEiFJ6tsgZgzn\nAosi4mrgPuDJMa9fwupT3CRJ08CkE9zWJCJeDWyWmVdGxPbA5Zm5Xasf0ssEt1G6OCUejzWpWY9e\nXazJmhLcBjFjuBtYHBEfodw/OHYAnyFJGpDWG0NmLgP2afu4kqThmDbffJYkTQ82BklSxcYgSarY\nGCRJFRuDJKliY5AkVabNstv9GFaC23RhkpykQXLGIEmq2BgkSRUbgySpMtB7DE2a2xcoSWwbA39C\nCfLZDtgQOCMzv9Kkt40kvC0G9gWe2+x3amYuGuQ4JUnPGvTN5/cA92bmWyPiZcCBwCOZ+baImAPc\nFBFXNfuOJLwdATwvM/dtVme9BFg04HGuVyaSMmUSVS9rUrMevaxJMejGEMBlAJl5a0QcA3y7ef5E\nRNxOmRXAswlvALc0j9MqwW26GG9p4C4uHzwea1KzHr26WJNhJriNdgfwSoCIeDFwCLBH83wOMB+4\np9l35aj3meAmSVNk0I3hLODFEbEUOB94PbBFRFwLLAE+lpkPDXgMkqR10HqC2xQxwW2ULk6Jx2NN\natajVxdrsqYEN/+5qiSpYmOQJFVsDJKkio1BklSxMUiSKjYGSVLFxiBJqtgYJEkVE9wkaT01qDRH\nZwySpIqNQZJUsTFIkiqt3GNoktrOB7amZCjsCdzFs6lsbwHOAZ4PbAmck5kLI+K9wNspS25fm5l/\nFhEHAScCTwP3Aodn5kokSUPR1s3no4F7MvPgiNgBuI3SGEZS2X4buCAzL46IrYGlwELgHcAfZeYN\nEXFMRGxEyWz4m8y8ICIOBzYHftrSOCVpxhhU4lxbjWFH4HKAzLwzIh5uto+ksi0DjmtmA49T8p+h\nNIYTIuJU4HpgFvCnwAebtLc7gK+3NEZJmlH6XSZ80AlutwK7AUTEdpTLRfBsKtsJwPWZeRhwIaUB\nALwLeE9m7gX8b+DVlNnHR5tts4DfbWmMkqQJaGvGcC6wKCKuBu4Dnhzz+iXAwog4FHgUWBERmwA/\nBL7fzDAeAL5HuXT0zxHxKPAE8M2WxihJmoBWEtwi4tXAZpl5ZURsD1yemdv1feCJM8FtlC4mUY3H\nmtSsR68u1mRNCW5tzRjuBhZHxEco9w+Obem4kqQha6UxZOYyYJ82jiVJmlp+wU2SVLExSJIqrdx8\nliTNHM4YJEkVG4MkqWJjkCRVbAySpIqNQZJUsTFIkio2BklSpa21koYuIjYAPgv8L2A58M7M/H9T\nO6p2RcQuwKmZuXdE/CawCFhFWeb82Mxc2axP9UZgBXBcZt7Yxr7DPM+JiIiNgfOA/wlsAvwlcDsd\nrUlEbEhJRQzgGUq2ySw6Wo8REfEC4AfAAso5LKLD9Zis9XnGcCAwOzN3A04CTp/i8bQqIj4AfB6Y\n3Ww6Azg5M/eg/AI4oEnG2wvYBXgr8Jk29h30uU3SYcCjzTj3Az5Nt2uyP0Bmvgb4MGXMXa7HyB8P\nZwG/bDZ1uh79WJ8bw+48mxp3A/CKqR1O6/4dOGjU85dTIlEBLgNeR6nBlZm5KjN/DGwUEfNa2Hc6\nuhD4i1HPV9DhmmTm1ymhVgDbAA/S4Xo0PgV8DvhJ87zr9Zi09bkxbA78bNTzZ5rM6BkhM/8ReHrU\nplmZObJ+yRPA8+itwcj2fveddjLz55n5RETMAS4CTsaarIiILwJ/T6lJZ+sREUcAD2fmFaM2d7Ye\n/VqfG8PjwOjA0g0yc8VUDWYIRl/DnAP8lN4ajGzvd99pKSJeCHwX+D+Z+WWsCZn5duAllPsNm456\nqWv1OBJYEBFLgJ2B84EXjHq9a/Xoy/rcGP4FeANAROxKiQmdyW6OiL2bn/cDrqHUYN+I2CAiXkRp\njo+0sO+0ExG/BlwJnJiZ5zWbO1uTiHhbRHywefoLyi+rf+1qPTJzz8zcKzP3Bm4BDgcu62o9+rU+\nX3r5GuUvhOsoN4DeMcXjGbTjgXMi4jnAHcBFmflMRFwDXE9p8se2se/QzmjdfAiYC/xFRIzca3g/\ncGZHa3Ix8IUmZ31j4DjKWLv838hYXf9/ZtJcdluSVFmfLyVJkgbAxiBJqtgYJEkVG4MkqWJjkCRV\nbAySpIqNQZJU+f8Veu1jXDiP0QAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x11a3a8ad0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "df16.drop('fk',axis=1,inplace=True)\n",
    "df16.c2015.value_counts().plot(kind='barh')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 139,
   "metadata": {
    "collapsed": false,
    "deletable": true,
    "editable": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "soil           71\n",
       "cell_id         0\n",
       "gid             0\n",
       "c2016           0\n",
       "c2015       75985\n",
       "c2014      135702\n",
       "c2013      178223\n",
       "c2012      203009\n",
       "c2011      229724\n",
       "c2010      252030\n",
       "c2009      278715\n",
       "dtype: int64"
      ]
     },
     "execution_count": 139,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df16.isnull().sum()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 207,
   "metadata": {
    "collapsed": false,
    "deletable": true,
    "editable": true,
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Percent of parcels changed from last year 9.73%\n"
     ]
    }
   ],
   "source": [
    "print \"Percent of parcels changed from last year {0:.2f}%\".format(refDf.c2015.isnull().sum() *100.0/ len(refDf))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 153,
   "metadata": {
    "collapsed": false,
    "deletable": true,
    "editable": true
   },
   "outputs": [],
   "source": [
    "nna_ref = df16.dropna() \n",
    "potatos = nna_ref[nna_ref.c2016.apply(lambda x: True if \"potato\" in x.lower() else False)]\n",
    "\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 144,
   "metadata": {
    "collapsed": false,
    "deletable": true,
    "editable": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Zand              9416\n",
       "Lichte zavel      3503\n",
       "Zware zavel       2444\n",
       "Lichte klei       1346\n",
       "Veen               680\n",
       "Moerig op zand     507\n",
       "Leem               420\n",
       "Zware klei         251\n",
       "Bebouwing, enz      69\n",
       "Water                7\n",
       "Name: soil, dtype: int64"
      ]
     },
     "execution_count": 144,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "potatos.soil.value_counts()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "deletable": true,
    "editable": true
   },
   "source": [
    "#ToDO check those on water  "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 156,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "potatos = potatos.applymap(lambda x: 1 if x==\"potato\" else 0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 159,
   "metadata": {
    "collapsed": false,
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "2    6333\n",
       "3    5259\n",
       "1    3753\n",
       "4    2984\n",
       "5     282\n",
       "6      26\n",
       "8       4\n",
       "7       2\n",
       "dtype: int64"
      ]
     },
     "execution_count": 159,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "potatos.sum(axis=1).value_counts()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 226,
   "metadata": {
    "collapsed": false,
    "deletable": true,
    "editable": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "MultiIndex(levels=[[u'Bebouwing, enz', u'Leem', u'Lichte klei', u'Lichte zavel', u'Moerig op zand', u'Veen', u'Water', u'Zand', u'Zware klei', u'Zware zavel'], [u'Aardappelen, consumptie', u'Aardappelen, poot NAK', u'Aardappelen, zetmeel', u'Bieten, suiker-', u'Grasland, blijvend', u'Grasland, natuurlijk. Areaal met een natuurbeheertype dat overwegend voor landbouwactiviteiten-GLB wordt gebruikt', u'Grasland, natuurlijk. Hoofdfunctie landbouw.', u'Grasland, natuurlijk. Hoofdfunctie natuur.', u'Grasland, tijdelijk', u'Maïs, snij-', u'Tarwe, winter-']],\n",
       "           labels=[[0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 2, 2, 2, 2, 2, 3, 3, 3, 3, 3, 4, 4, 4, 4, 4, 5, 5, 5, 5, 5, 6, 6, 6, 6, 6, 7, 7, 7, 7, 7, 8, 8, 8, 8, 8, 9, 9, 9, 9, 9], [4, 8, 6, 9, 7, 4, 9, 10, 8, 3, 4, 8, 10, 9, 0, 4, 8, 9, 10, 1, 4, 8, 9, 6, 2, 4, 6, 8, 5, 9, 4, 6, 5, 7, 8, 4, 9, 8, 0, 2, 4, 8, 9, 10, 6, 4, 8, 9, 10, 0]],\n",
       "           names=[u'soil', u'c2016'])"
      ]
     },
     "execution_count": 226,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_agg  = nna_ref.groupby(['soil','c2016']).agg({'gid':'count'})\n",
    "g = df_agg['gid'].groupby(level=0, group_keys=False)\n",
    "res = g.nlargest(5)\n",
    "# res.plot(kind='barh',figsize=(5,10))\n",
    "res.index\n",
    "#ToDo add subplots based on soil type"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "deletable": true,
    "editable": true
   },
   "source": [
    "### Prediction"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 292,
   "metadata": {
    "collapsed": false,
    "deletable": true,
    "editable": true
   },
   "outputs": [],
   "source": [
    "## TEST ## \n",
    "#rename df columns\n",
    "\n",
    "# d = {}\n",
    "# cols = refDf.ix[:1,\"c2016\":].columns.values\n",
    "# for i in range(len (cols[1:])+1):\n",
    "#     d[cols[i]]= str(i)\n",
    "    \n",
    "# d[cols[0]]= \"y\"\n",
    "# tdf = refDf.copy()\n",
    "# tdf.rename(columns=d)\n",
    "\n",
    "\n",
    "\n",
    "# rename columns dic\n",
    "# d = {}\n",
    "# for di in range (1,5):\n",
    "#     d[str(di)] = []\n",
    "# d['y']= []\n",
    "# mldf = pd.DataFrame(d)\n",
    "# mldf"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "metadata": {
    "collapsed": false,
    "deletable": true,
    "editable": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "adding target  2015\n",
      "adding target  2016\n"
     ]
    }
   ],
   "source": [
    "# USAGE: pd.concat([df10, append_nils_for_previous_years(2010,len(df10))], axis=1)\n",
    "\n",
    "def append_nils_for_previous_years(year, length,window=4): \n",
    "    nullDF = pd.DataFrame()\n",
    "\n",
    "    for y in range (2008,year-window, -1):\n",
    "        nullDF[(\"c\"+str(y))] = [np.NaN]*length\n",
    "        \n",
    "        \n",
    "    return nullDF    \n",
    "\n",
    "\n",
    "def create_dataset(start,end,window=4, prune_grass=False): #min 2010, max 2016\n",
    "#     y = []\n",
    "#     X = []\n",
    "\n",
    "    #create base dataframe and append results of each loop\n",
    "    #we rename columns to 1:window so that columns across varying windows can be concatenated\n",
    "    d = {}\n",
    "    for di in range (1,window+1):\n",
    "        d[str(di)] = []\n",
    "    d['y']= []\n",
    "    mldf = pd.DataFrame(d)\n",
    "    \n",
    "    \n",
    "    for year in range(start,end+1):\n",
    "        join_df = get_backward_join_of_parcels(year,max(year-window,2009),True)\n",
    "       \n",
    "        # one way is to pad from 2003 to 2008 with zeros. \n",
    "        join_df = pd.concat([join_df, append_nils_for_previous_years(year,len(join_df),window)], axis=1)\n",
    "#         print 'after null', join_df.head(1)\n",
    "        print \"adding target \", year \n",
    "        join_df = join_df[~join_df[(\"c\"+str(year-1))].isnull()] # prediction target is join of this year's data with that of next year \n",
    "#         X.extend(join_df.ix[:,(\"c\"+str(year-1)):].values)\n",
    "#         y.extend(join_df.ix[:,(\"c\"+str(year))].values)\n",
    "        \n",
    "        #rename columns\n",
    "        \n",
    "        d = {}\n",
    "        cols = join_df.ix[:1,(\"c\"+str(year)):].columns.values\n",
    "        for i in range(len (cols[1:])+1):\n",
    "            d[cols[i]]= str(i)\n",
    "\n",
    "        d[cols[0]]= \"y\"\n",
    "        join_df.rename(columns=d,inplace=True)        \n",
    "        join_df.drop(['gid','fk'],axis=1,inplace=True)\n",
    "        \n",
    "        #not working\n",
    "#         if prune_grass: #remove all parcels that have had value 'grass' across ALL the years\n",
    "#             join_df = join_df[join_df.apply(lambda row: False if \\\n",
    "#                                             (np.unique(row.values[0:window])==['grass'])[0] \\\n",
    "#                                             else True,axis=1)]\n",
    "\n",
    "        #append to existing dataset\n",
    "        mldf = pd.concat([mldf,join_df])\n",
    "        \n",
    "\n",
    "    return mldf\n",
    "\n",
    "t = create_dataset(2015,2016,6)      \n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {
    "collapsed": false,
    "deletable": true,
    "editable": true
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/aoliya/anaconda/envs/test_fiona5/lib/python2.7/site-packages/numpy/lib/arraysetops.py:200: FutureWarning: numpy not_equal will not check object identity in the future. The comparison did not return the same result as suggested by the identity (`is`)) and will change.\n",
      "  flag = np.concatenate(([True], aux[1:] != aux[:-1]))\n"
     ]
    }
   ],
   "source": [
    "\n",
    "\n",
    "tprime = t[t.apply(lambda row: False if (np.unique(row.values[0:5])==['grass'])[0] else True,axis=1)]\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "metadata": {
    "collapsed": false,
    "deletable": true,
    "editable": true,
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "1740067"
      ]
     },
     "execution_count": 53,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(tprime)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "metadata": {
    "collapsed": false,
    "deletable": true,
    "editable": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "2788383"
      ]
     },
     "execution_count": 54,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(t)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "metadata": {
    "collapsed": false,
    "deletable": true,
    "editable": true
   },
   "outputs": [],
   "source": [
    "from sklearn.ensemble import RandomForestClassifier\n",
    "from sklearn.metrics import classification_report\n",
    "from sklearn import preprocessing \n",
    "\n",
    "def train_validate_predict(df,weights=None):\n",
    "    \n",
    "#   it seems decision tree/random forest can still work ok with integers   \n",
    "#     le = preprocessing.LabelEncoder()\n",
    "#     le.fit(np.unique(df.values.reshapre(1,-1)))\n",
    "#     df = df.apply(le.transform, axis=1)\n",
    "#     df = pd.get_dummies(df,dummy_na=True)\n",
    "    \n",
    "    df = df.sample(frac=1) #randomize    \n",
    "    traindf = df.iloc[0:int(0.7*len(df))]\n",
    "    testdf = df.iloc[int(0.7*len(df)):]\n",
    "\n",
    "    X_train, X_valid, y_train, y_valid = train_test_split(traindf.ix[:,0:len(df.columns)-1],\\\n",
    "                                                          traindf.ix[:,len(df.columns)-1],\\\n",
    "                                                          test_size=0.33, random_state=19,\\\n",
    "                                                          stratify=traindf.ix[:,len(df.columns)-1])\n",
    "\n",
    "\n",
    "    X_test = testdf.ix[:,0:len(df.columns)-1]\n",
    "    y_test = testdf.ix[:,len(df.columns)-1]\n",
    "\n",
    "    clf = RandomForestClassifier(n_estimators=100,n_jobs=-1,class_weight=weights, random_state=557) #set class weight for potatos \n",
    "    clf.fit(X_train,y_train)\n",
    "\n",
    "    print \"Results on validation set:\\n\"\n",
    "    print(classification_report(y_valid, clf.predict(X_valid)))\n",
    "\n",
    "    # print \"And results on test set:\\n\"\n",
    "    # print(classification_report(X_test,y_test ))\n",
    "\n",
    "    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 233,
   "metadata": {
    "collapsed": false,
    "deletable": true,
    "editable": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "['beans' 'corn' 'flower' 'grain' 'grass' 'onion' 'others' 'potato' 'seed'\n",
      " 'tree' 'veggies' 'waste']\n"
     ]
    }
   ],
   "source": [
    "from sklearn import preprocessing \n",
    "\n",
    "t_simple = t.drop([\"cell_id\",\"soil\"],axis=1)\n",
    "t_simple.fillna(\"others\",inplace=True)\n",
    "\n",
    "le = preprocessing.LabelEncoder()\n",
    "le.fit(np.unique(t_simple.values.reshape(1,-1)))\n",
    "t_simple = t_simple.apply(le.transform)\n",
    "t_simple.reset_index().drop('index',axis=1,inplace=True)\n",
    "\n",
    "print le.classes_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 235,
   "metadata": {
    "collapsed": false,
    "deletable": true,
    "editable": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Results on validation set:\n",
      "\n",
      "             precision    recall  f1-score   support\n",
      "\n",
      "          0       0.01      0.21      0.03       979\n",
      "          1       0.75      0.69      0.72     80570\n",
      "          2       0.46      0.44      0.45     13131\n",
      "          3       0.53      0.49      0.51     34225\n",
      "          4       0.97      0.94      0.95    426487\n",
      "          5       0.45      0.37      0.41     22415\n",
      "          6       0.02      0.31      0.03       423\n",
      "          7       0.41      0.46      0.44     27310\n",
      "          8       0.35      0.50      0.41      9304\n",
      "          9       0.72      0.78      0.75     11052\n",
      "         10       0.31      0.31      0.31      5423\n",
      "         11       0.84      0.80      0.82     12748\n",
      "\n",
      "avg / total       0.84      0.81      0.83    644067\n",
      "\n"
     ]
    }
   ],
   "source": [
    "train_validate_predict(t_simple,\"balanced\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 112,
   "metadata": {
    "collapsed": false,
    "deletable": true,
    "editable": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Results on validation set:\n",
      "\n",
      "             precision    recall  f1-score   support\n",
      "\n",
      "          0       0.21      0.01      0.02       705\n",
      "          1       0.38      0.05      0.08       174\n",
      "          2       0.00      0.00      0.00       474\n",
      "          3       0.69      0.80      0.74     61692\n",
      "          4       0.68      0.52      0.59      6571\n",
      "          5       0.49      0.62      0.55     25989\n",
      "          6       0.96      0.96      0.96    316314\n",
      "          8       0.51      0.37      0.43     17549\n",
      "          9       0.57      0.38      0.45     15814\n",
      "         10       0.54      0.40      0.46     20231\n",
      "         11       0.36      0.21      0.26      1938\n",
      "         12       0.61      0.89      0.73      5575\n",
      "         13       0.85      0.75      0.79      8258\n",
      "\n",
      "avg / total       0.84      0.84      0.84    481284\n",
      "\n"
     ]
    }
   ],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "deletable": true,
    "editable": true
   },
   "source": [
    "#### with soil"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 148,
   "metadata": {
    "collapsed": false,
    "deletable": true,
    "editable": true
   },
   "outputs": [],
   "source": [
    "tsoil = t.drop('cell_id',axis=1)\n",
    "tsoil = tsoil.fillna('others')\n",
    "ls = preprocessing.LabelEncoder()\n",
    "ls.fit(np.unique(tsoil.values.reshape(1,-1)))\n",
    "tsoil = tsoil.apply(ls.transform)\n",
    "tsoil.reset_index().drop('index',axis=1,inplace=True)\n",
    "\n",
    "print ls.classes_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 152,
   "metadata": {
    "collapsed": false,
    "deletable": true,
    "editable": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Results on validation set:\n",
      "\n",
      "             precision    recall  f1-score   support\n",
      "\n",
      "         12       0.11      0.01      0.02       666\n",
      "         13       0.29      0.05      0.08       177\n",
      "         14       0.69      0.80      0.74     61751\n",
      "         15       0.66      0.50      0.57      6554\n",
      "         16       0.52      0.61      0.56     25955\n",
      "         17       0.96      0.96      0.96    316234\n",
      "         18       0.49      0.39      0.43     17517\n",
      "         19       0.58      0.37      0.45     16370\n",
      "         20       0.54      0.44      0.48     20324\n",
      "         21       0.35      0.21      0.26      1954\n",
      "         22       0.60      0.88      0.71      5528\n",
      "         23       0.84      0.75      0.79      8254\n",
      "\n",
      "avg / total       0.84      0.84      0.84    481284\n",
      "\n"
     ]
    }
   ],
   "source": [
    "train_validate_predict(tsoil)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 95,
   "metadata": {
    "collapsed": false,
    "deletable": true,
    "editable": true
   },
   "outputs": [],
   "source": [
    "tzegond = t[t.apply(lambda row: True if ('potato' in row.values[0:4])  else False,axis=1)]\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {
    "collapsed": false,
    "deletable": true,
    "editable": true,
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "from sklearn import preprocessing \n",
    "\n",
    "\n",
    "#with cell id \n",
    "nona = tprime[~tprime.soil.isnull()]\n",
    "fishml = nona.drop('cell_id',axis=1)\n",
    "fishml = fishml.fillna('others')\n",
    "\n",
    "ls = preprocessing.LabelEncoder()\n",
    "ls.fit(np.unique(fishml.values.reshape(1,-1)))\n",
    "fishml = fishml.apply(ls.transform)\n",
    "\n",
    "fishml.insert(0,\"cell_id\",nona.cell_id)\n",
    "fishml.reset_index().drop('index',axis=1,inplace=True)\n",
    "\n",
    "\n",
    "# train_validate_predict(fishml)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 98,
   "metadata": {
    "collapsed": false,
    "deletable": true,
    "editable": true
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>cell_id</th>\n",
       "      <th>1</th>\n",
       "      <th>2</th>\n",
       "      <th>3</th>\n",
       "      <th>4</th>\n",
       "      <th>soil</th>\n",
       "      <th>y</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>51</th>\n",
       "      <td>530.0</td>\n",
       "      <td>18</td>\n",
       "      <td>13</td>\n",
       "      <td>17</td>\n",
       "      <td>15</td>\n",
       "      <td>2</td>\n",
       "      <td>14</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>55</th>\n",
       "      <td>772.0</td>\n",
       "      <td>17</td>\n",
       "      <td>13</td>\n",
       "      <td>13</td>\n",
       "      <td>17</td>\n",
       "      <td>9</td>\n",
       "      <td>14</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>102</th>\n",
       "      <td>583.0</td>\n",
       "      <td>17</td>\n",
       "      <td>12</td>\n",
       "      <td>18</td>\n",
       "      <td>15</td>\n",
       "      <td>3</td>\n",
       "      <td>14</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>110</th>\n",
       "      <td>686.0</td>\n",
       "      <td>14</td>\n",
       "      <td>17</td>\n",
       "      <td>14</td>\n",
       "      <td>14</td>\n",
       "      <td>2</td>\n",
       "      <td>14</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>112</th>\n",
       "      <td>686.0</td>\n",
       "      <td>14</td>\n",
       "      <td>14</td>\n",
       "      <td>17</td>\n",
       "      <td>14</td>\n",
       "      <td>7</td>\n",
       "      <td>14</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>137</th>\n",
       "      <td>611.0</td>\n",
       "      <td>17</td>\n",
       "      <td>18</td>\n",
       "      <td>15</td>\n",
       "      <td>17</td>\n",
       "      <td>4</td>\n",
       "      <td>14</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>179</th>\n",
       "      <td>618.0</td>\n",
       "      <td>14</td>\n",
       "      <td>14</td>\n",
       "      <td>14</td>\n",
       "      <td>17</td>\n",
       "      <td>7</td>\n",
       "      <td>14</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>182</th>\n",
       "      <td>618.0</td>\n",
       "      <td>11</td>\n",
       "      <td>17</td>\n",
       "      <td>11</td>\n",
       "      <td>11</td>\n",
       "      <td>7</td>\n",
       "      <td>14</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>184</th>\n",
       "      <td>618.0</td>\n",
       "      <td>11</td>\n",
       "      <td>17</td>\n",
       "      <td>11</td>\n",
       "      <td>11</td>\n",
       "      <td>7</td>\n",
       "      <td>14</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>375</th>\n",
       "      <td>618.0</td>\n",
       "      <td>14</td>\n",
       "      <td>13</td>\n",
       "      <td>17</td>\n",
       "      <td>14</td>\n",
       "      <td>7</td>\n",
       "      <td>14</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>399</th>\n",
       "      <td>563.0</td>\n",
       "      <td>14</td>\n",
       "      <td>17</td>\n",
       "      <td>12</td>\n",
       "      <td>11</td>\n",
       "      <td>7</td>\n",
       "      <td>14</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>401</th>\n",
       "      <td>374.0</td>\n",
       "      <td>14</td>\n",
       "      <td>17</td>\n",
       "      <td>11</td>\n",
       "      <td>14</td>\n",
       "      <td>7</td>\n",
       "      <td>14</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>448</th>\n",
       "      <td>588.0</td>\n",
       "      <td>14</td>\n",
       "      <td>14</td>\n",
       "      <td>11</td>\n",
       "      <td>17</td>\n",
       "      <td>7</td>\n",
       "      <td>14</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>486</th>\n",
       "      <td>618.0</td>\n",
       "      <td>14</td>\n",
       "      <td>17</td>\n",
       "      <td>17</td>\n",
       "      <td>14</td>\n",
       "      <td>7</td>\n",
       "      <td>14</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>487</th>\n",
       "      <td>618.0</td>\n",
       "      <td>17</td>\n",
       "      <td>14</td>\n",
       "      <td>14</td>\n",
       "      <td>14</td>\n",
       "      <td>7</td>\n",
       "      <td>14</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>491</th>\n",
       "      <td>230.0</td>\n",
       "      <td>14</td>\n",
       "      <td>17</td>\n",
       "      <td>11</td>\n",
       "      <td>11</td>\n",
       "      <td>7</td>\n",
       "      <td>14</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>501</th>\n",
       "      <td>637.0</td>\n",
       "      <td>14</td>\n",
       "      <td>14</td>\n",
       "      <td>14</td>\n",
       "      <td>17</td>\n",
       "      <td>7</td>\n",
       "      <td>14</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>502</th>\n",
       "      <td>637.0</td>\n",
       "      <td>14</td>\n",
       "      <td>14</td>\n",
       "      <td>14</td>\n",
       "      <td>17</td>\n",
       "      <td>7</td>\n",
       "      <td>14</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>521</th>\n",
       "      <td>208.0</td>\n",
       "      <td>14</td>\n",
       "      <td>17</td>\n",
       "      <td>12</td>\n",
       "      <td>10</td>\n",
       "      <td>7</td>\n",
       "      <td>14</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>557</th>\n",
       "      <td>746.0</td>\n",
       "      <td>14</td>\n",
       "      <td>11</td>\n",
       "      <td>17</td>\n",
       "      <td>13</td>\n",
       "      <td>9</td>\n",
       "      <td>14</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>709</th>\n",
       "      <td>228.0</td>\n",
       "      <td>14</td>\n",
       "      <td>17</td>\n",
       "      <td>11</td>\n",
       "      <td>16</td>\n",
       "      <td>7</td>\n",
       "      <td>14</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>777</th>\n",
       "      <td>613.0</td>\n",
       "      <td>14</td>\n",
       "      <td>17</td>\n",
       "      <td>14</td>\n",
       "      <td>14</td>\n",
       "      <td>5</td>\n",
       "      <td>14</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>910</th>\n",
       "      <td>508.0</td>\n",
       "      <td>14</td>\n",
       "      <td>11</td>\n",
       "      <td>11</td>\n",
       "      <td>17</td>\n",
       "      <td>7</td>\n",
       "      <td>14</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>930</th>\n",
       "      <td>801.0</td>\n",
       "      <td>17</td>\n",
       "      <td>13</td>\n",
       "      <td>13</td>\n",
       "      <td>17</td>\n",
       "      <td>3</td>\n",
       "      <td>14</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1013</th>\n",
       "      <td>202.0</td>\n",
       "      <td>17</td>\n",
       "      <td>11</td>\n",
       "      <td>16</td>\n",
       "      <td>16</td>\n",
       "      <td>7</td>\n",
       "      <td>14</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1020</th>\n",
       "      <td>509.0</td>\n",
       "      <td>14</td>\n",
       "      <td>17</td>\n",
       "      <td>11</td>\n",
       "      <td>11</td>\n",
       "      <td>5</td>\n",
       "      <td>14</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1077</th>\n",
       "      <td>563.0</td>\n",
       "      <td>14</td>\n",
       "      <td>14</td>\n",
       "      <td>14</td>\n",
       "      <td>17</td>\n",
       "      <td>3</td>\n",
       "      <td>14</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1106</th>\n",
       "      <td>203.0</td>\n",
       "      <td>11</td>\n",
       "      <td>11</td>\n",
       "      <td>17</td>\n",
       "      <td>15</td>\n",
       "      <td>7</td>\n",
       "      <td>14</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1178</th>\n",
       "      <td>801.0</td>\n",
       "      <td>14</td>\n",
       "      <td>17</td>\n",
       "      <td>13</td>\n",
       "      <td>15</td>\n",
       "      <td>3</td>\n",
       "      <td>14</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1332</th>\n",
       "      <td>501.0</td>\n",
       "      <td>15</td>\n",
       "      <td>14</td>\n",
       "      <td>17</td>\n",
       "      <td>16</td>\n",
       "      <td>2</td>\n",
       "      <td>14</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",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>760078</th>\n",
       "      <td>502.0</td>\n",
       "      <td>18</td>\n",
       "      <td>17</td>\n",
       "      <td>15</td>\n",
       "      <td>16</td>\n",
       "      <td>9</td>\n",
       "      <td>14</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>760095</th>\n",
       "      <td>228.0</td>\n",
       "      <td>17</td>\n",
       "      <td>16</td>\n",
       "      <td>16</td>\n",
       "      <td>16</td>\n",
       "      <td>7</td>\n",
       "      <td>14</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>760232</th>\n",
       "      <td>195.0</td>\n",
       "      <td>16</td>\n",
       "      <td>21</td>\n",
       "      <td>21</td>\n",
       "      <td>17</td>\n",
       "      <td>2</td>\n",
       "      <td>14</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>760377</th>\n",
       "      <td>774.0</td>\n",
       "      <td>14</td>\n",
       "      <td>17</td>\n",
       "      <td>15</td>\n",
       "      <td>11</td>\n",
       "      <td>3</td>\n",
       "      <td>14</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>760555</th>\n",
       "      <td>400.0</td>\n",
       "      <td>14</td>\n",
       "      <td>11</td>\n",
       "      <td>18</td>\n",
       "      <td>17</td>\n",
       "      <td>7</td>\n",
       "      <td>14</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>760625</th>\n",
       "      <td>588.0</td>\n",
       "      <td>14</td>\n",
       "      <td>17</td>\n",
       "      <td>17</td>\n",
       "      <td>14</td>\n",
       "      <td>7</td>\n",
       "      <td>14</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>760658</th>\n",
       "      <td>330.0</td>\n",
       "      <td>14</td>\n",
       "      <td>14</td>\n",
       "      <td>17</td>\n",
       "      <td>18</td>\n",
       "      <td>3</td>\n",
       "      <td>14</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>760843</th>\n",
       "      <td>286.0</td>\n",
       "      <td>14</td>\n",
       "      <td>14</td>\n",
       "      <td>17</td>\n",
       "      <td>11</td>\n",
       "      <td>7</td>\n",
       "      <td>14</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>761066</th>\n",
       "      <td>671.0</td>\n",
       "      <td>14</td>\n",
       "      <td>14</td>\n",
       "      <td>14</td>\n",
       "      <td>17</td>\n",
       "      <td>7</td>\n",
       "      <td>14</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>761080</th>\n",
       "      <td>564.0</td>\n",
       "      <td>17</td>\n",
       "      <td>11</td>\n",
       "      <td>17</td>\n",
       "      <td>15</td>\n",
       "      <td>7</td>\n",
       "      <td>14</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>761142</th>\n",
       "      <td>780.0</td>\n",
       "      <td>17</td>\n",
       "      <td>13</td>\n",
       "      <td>11</td>\n",
       "      <td>15</td>\n",
       "      <td>9</td>\n",
       "      <td>14</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>761161</th>\n",
       "      <td>779.0</td>\n",
       "      <td>17</td>\n",
       "      <td>16</td>\n",
       "      <td>16</td>\n",
       "      <td>16</td>\n",
       "      <td>9</td>\n",
       "      <td>14</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>761202</th>\n",
       "      <td>611.0</td>\n",
       "      <td>15</td>\n",
       "      <td>17</td>\n",
       "      <td>12</td>\n",
       "      <td>13</td>\n",
       "      <td>3</td>\n",
       "      <td>14</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>761204</th>\n",
       "      <td>611.0</td>\n",
       "      <td>15</td>\n",
       "      <td>17</td>\n",
       "      <td>12</td>\n",
       "      <td>13</td>\n",
       "      <td>3</td>\n",
       "      <td>14</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>761290</th>\n",
       "      <td>632.0</td>\n",
       "      <td>17</td>\n",
       "      <td>11</td>\n",
       "      <td>11</td>\n",
       "      <td>11</td>\n",
       "      <td>9</td>\n",
       "      <td>14</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>761315</th>\n",
       "      <td>674.0</td>\n",
       "      <td>11</td>\n",
       "      <td>17</td>\n",
       "      <td>11</td>\n",
       "      <td>11</td>\n",
       "      <td>7</td>\n",
       "      <td>14</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>761394</th>\n",
       "      <td>225.0</td>\n",
       "      <td>14</td>\n",
       "      <td>14</td>\n",
       "      <td>17</td>\n",
       "      <td>11</td>\n",
       "      <td>7</td>\n",
       "      <td>14</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>761495</th>\n",
       "      <td>638.0</td>\n",
       "      <td>17</td>\n",
       "      <td>12</td>\n",
       "      <td>16</td>\n",
       "      <td>16</td>\n",
       "      <td>4</td>\n",
       "      <td>14</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>761526</th>\n",
       "      <td>234.0</td>\n",
       "      <td>17</td>\n",
       "      <td>18</td>\n",
       "      <td>12</td>\n",
       "      <td>16</td>\n",
       "      <td>7</td>\n",
       "      <td>14</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>761676</th>\n",
       "      <td>671.0</td>\n",
       "      <td>14</td>\n",
       "      <td>17</td>\n",
       "      <td>11</td>\n",
       "      <td>11</td>\n",
       "      <td>7</td>\n",
       "      <td>14</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>761862</th>\n",
       "      <td>610.0</td>\n",
       "      <td>17</td>\n",
       "      <td>16</td>\n",
       "      <td>16</td>\n",
       "      <td>16</td>\n",
       "      <td>3</td>\n",
       "      <td>14</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>762067</th>\n",
       "      <td>288.0</td>\n",
       "      <td>14</td>\n",
       "      <td>14</td>\n",
       "      <td>11</td>\n",
       "      <td>17</td>\n",
       "      <td>7</td>\n",
       "      <td>14</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>762221</th>\n",
       "      <td>585.0</td>\n",
       "      <td>14</td>\n",
       "      <td>11</td>\n",
       "      <td>11</td>\n",
       "      <td>17</td>\n",
       "      <td>7</td>\n",
       "      <td>14</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>762286</th>\n",
       "      <td>398.0</td>\n",
       "      <td>11</td>\n",
       "      <td>17</td>\n",
       "      <td>11</td>\n",
       "      <td>11</td>\n",
       "      <td>9</td>\n",
       "      <td>14</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>762348</th>\n",
       "      <td>669.0</td>\n",
       "      <td>14</td>\n",
       "      <td>14</td>\n",
       "      <td>11</td>\n",
       "      <td>17</td>\n",
       "      <td>7</td>\n",
       "      <td>14</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>762349</th>\n",
       "      <td>644.0</td>\n",
       "      <td>14</td>\n",
       "      <td>17</td>\n",
       "      <td>14</td>\n",
       "      <td>14</td>\n",
       "      <td>7</td>\n",
       "      <td>14</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>762447</th>\n",
       "      <td>674.0</td>\n",
       "      <td>17</td>\n",
       "      <td>13</td>\n",
       "      <td>16</td>\n",
       "      <td>16</td>\n",
       "      <td>7</td>\n",
       "      <td>14</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>762449</th>\n",
       "      <td>674.0</td>\n",
       "      <td>14</td>\n",
       "      <td>14</td>\n",
       "      <td>17</td>\n",
       "      <td>11</td>\n",
       "      <td>7</td>\n",
       "      <td>14</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>762450</th>\n",
       "      <td>674.0</td>\n",
       "      <td>14</td>\n",
       "      <td>14</td>\n",
       "      <td>17</td>\n",
       "      <td>16</td>\n",
       "      <td>7</td>\n",
       "      <td>14</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>762729</th>\n",
       "      <td>773.0</td>\n",
       "      <td>14</td>\n",
       "      <td>17</td>\n",
       "      <td>14</td>\n",
       "      <td>14</td>\n",
       "      <td>3</td>\n",
       "      <td>14</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>9946 rows × 7 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "        cell_id   1   2   3   4  soil   y\n",
       "51        530.0  18  13  17  15     2  14\n",
       "55        772.0  17  13  13  17     9  14\n",
       "102       583.0  17  12  18  15     3  14\n",
       "110       686.0  14  17  14  14     2  14\n",
       "112       686.0  14  14  17  14     7  14\n",
       "137       611.0  17  18  15  17     4  14\n",
       "179       618.0  14  14  14  17     7  14\n",
       "182       618.0  11  17  11  11     7  14\n",
       "184       618.0  11  17  11  11     7  14\n",
       "375       618.0  14  13  17  14     7  14\n",
       "399       563.0  14  17  12  11     7  14\n",
       "401       374.0  14  17  11  14     7  14\n",
       "448       588.0  14  14  11  17     7  14\n",
       "486       618.0  14  17  17  14     7  14\n",
       "487       618.0  17  14  14  14     7  14\n",
       "491       230.0  14  17  11  11     7  14\n",
       "501       637.0  14  14  14  17     7  14\n",
       "502       637.0  14  14  14  17     7  14\n",
       "521       208.0  14  17  12  10     7  14\n",
       "557       746.0  14  11  17  13     9  14\n",
       "709       228.0  14  17  11  16     7  14\n",
       "777       613.0  14  17  14  14     5  14\n",
       "910       508.0  14  11  11  17     7  14\n",
       "930       801.0  17  13  13  17     3  14\n",
       "1013      202.0  17  11  16  16     7  14\n",
       "1020      509.0  14  17  11  11     5  14\n",
       "1077      563.0  14  14  14  17     3  14\n",
       "1106      203.0  11  11  17  15     7  14\n",
       "1178      801.0  14  17  13  15     3  14\n",
       "1332      501.0  15  14  17  16     2  14\n",
       "...         ...  ..  ..  ..  ..   ...  ..\n",
       "760078    502.0  18  17  15  16     9  14\n",
       "760095    228.0  17  16  16  16     7  14\n",
       "760232    195.0  16  21  21  17     2  14\n",
       "760377    774.0  14  17  15  11     3  14\n",
       "760555    400.0  14  11  18  17     7  14\n",
       "760625    588.0  14  17  17  14     7  14\n",
       "760658    330.0  14  14  17  18     3  14\n",
       "760843    286.0  14  14  17  11     7  14\n",
       "761066    671.0  14  14  14  17     7  14\n",
       "761080    564.0  17  11  17  15     7  14\n",
       "761142    780.0  17  13  11  15     9  14\n",
       "761161    779.0  17  16  16  16     9  14\n",
       "761202    611.0  15  17  12  13     3  14\n",
       "761204    611.0  15  17  12  13     3  14\n",
       "761290    632.0  17  11  11  11     9  14\n",
       "761315    674.0  11  17  11  11     7  14\n",
       "761394    225.0  14  14  17  11     7  14\n",
       "761495    638.0  17  12  16  16     4  14\n",
       "761526    234.0  17  18  12  16     7  14\n",
       "761676    671.0  14  17  11  11     7  14\n",
       "761862    610.0  17  16  16  16     3  14\n",
       "762067    288.0  14  14  11  17     7  14\n",
       "762221    585.0  14  11  11  17     7  14\n",
       "762286    398.0  11  17  11  11     9  14\n",
       "762348    669.0  14  14  11  17     7  14\n",
       "762349    644.0  14  17  14  14     7  14\n",
       "762447    674.0  17  13  16  16     7  14\n",
       "762449    674.0  14  14  17  11     7  14\n",
       "762450    674.0  14  14  17  16     7  14\n",
       "762729    773.0  14  17  14  14     3  14\n",
       "\n",
       "[9946 rows x 7 columns]"
      ]
     },
     "execution_count": 98,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "fishml[fishml.y==14]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false,
    "deletable": true,
    "editable": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x11f963c50>"
      ]
     },
     "execution_count": 131,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "fishml.plot(kind='line')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {
    "collapsed": false,
    "deletable": true,
    "editable": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "22 ['Bebouwing, enz' 'Leem' 'Lichte klei' 'Lichte zavel' 'Moerig op zand'\n",
      " 'Veen' 'Water' 'Zand' 'Zware klei' 'Zware zavel' 'beans' 'corn' 'flower'\n",
      " 'grain' 'grass' 'onion' 'others' 'potato' 'seed' 'tree' 'veggies' 'waste']\n"
     ]
    }
   ],
   "source": [
    "print len(ls.classes_),ls.classes_\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "deletable": true,
    "editable": true
   },
   "source": [
    "##### changing class weights"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "metadata": {
    "collapsed": false,
    "deletable": true,
    "editable": true,
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Results on validation set:\n",
      "\n",
      "             precision    recall  f1-score   support\n",
      "\n",
      "         10       0.12      0.06      0.08       961\n",
      "         11       0.70      0.78      0.74     76754\n",
      "         12       0.56      0.45      0.50     12575\n",
      "         13       0.53      0.57      0.55     33974\n",
      "         14       0.93      0.92      0.92    192408\n",
      "         15       0.43      0.38      0.40     21798\n",
      "         16       0.29      0.15      0.20       400\n",
      "         17       0.53      0.49      0.51     25836\n",
      "         18       0.42      0.39      0.40      9129\n",
      "         19       0.72      0.77      0.74     10824\n",
      "         20       0.44      0.32      0.37      5199\n",
      "         21       0.85      0.79      0.82     12058\n",
      "\n",
      "avg / total       0.76      0.76      0.76    401916\n",
      "\n"
     ]
    }
   ],
   "source": [
    "train_validate_predict(fishml)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 101,
   "metadata": {
    "collapsed": false,
    "deletable": true,
    "editable": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Results on validation set:\n",
      "\n",
      "             precision    recall  f1-score   support\n",
      "\n",
      "         10       0.11      0.05      0.07       103\n",
      "         11       0.55      0.51      0.53      3666\n",
      "         12       0.27      0.17      0.21       580\n",
      "         13       0.55      0.55      0.55      4788\n",
      "         14       0.67      0.60      0.63      2276\n",
      "         15       0.37      0.32      0.34      2829\n",
      "         16       0.00      0.00      0.00        39\n",
      "         17       0.52      0.73      0.61      3860\n",
      "         18       0.30      0.24      0.27      1313\n",
      "         19       0.45      0.37      0.41        84\n",
      "         21       0.45      0.29      0.35       185\n",
      "\n",
      "avg / total       0.50      0.51      0.50     19723\n",
      "\n"
     ]
    }
   ],
   "source": [
    "\n",
    "train_validate_predict(fishml,{17:100})"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false,
    "deletable": true,
    "editable": true
   },
   "outputs": [],
   "source": [
    "h2o.init()\n",
    "# mlset = np.concatenate((np.array(X),np.array(y).reshape(-1,1)),axis=1)\n",
    "hdf = h2o.H2OFrame.from_python(t)#,X.drop(['geom','center',],axis=1))\n",
    "\n",
    "print hdf.tail(1)\n",
    "gbm_classifier = H2OGradientBoostingEstimator(distribution=\"multinomial\", \\\n",
    "                                              ntrees=10, max_depth=5, min_rows=2, learn_rate=0.2 , seed=577,balance_classes=True)\n",
    "\n",
    "gbm_classifier.train(x=range(0,hdf.ncol), y=hdf.ncol-1, training_frame=hdf)\n",
    "\n",
    "gbm_classifier"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 313,
   "metadata": {
    "collapsed": false,
    "deletable": true,
    "editable": true,
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model Summary: \n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div style=\"overflow:auto\"><table style=\"width:50%\"><tr><td><b></b></td>\n",
       "<td><b>number_of_trees</b></td>\n",
       "<td><b>number_of_internal_trees</b></td>\n",
       "<td><b>model_size_in_bytes</b></td>\n",
       "<td><b>min_depth</b></td>\n",
       "<td><b>max_depth</b></td>\n",
       "<td><b>mean_depth</b></td>\n",
       "<td><b>min_leaves</b></td>\n",
       "<td><b>max_leaves</b></td>\n",
       "<td><b>mean_leaves</b></td></tr>\n",
       "<tr><td></td>\n",
       "<td>5.0</td>\n",
       "<td>60.0</td>\n",
       "<td>20465.0</td>\n",
       "<td>2.0</td>\n",
       "<td>2.0</td>\n",
       "<td>2.0</td>\n",
       "<td>4.0</td>\n",
       "<td>4.0</td>\n",
       "<td>4.0</td></tr></table></div>"
      ],
      "text/plain": [
       "    number_of_trees    number_of_internal_trees    model_size_in_bytes    min_depth    max_depth    mean_depth    min_leaves    max_leaves    mean_leaves\n",
       "--  -----------------  --------------------------  ---------------------  -----------  -----------  ------------  ------------  ------------  -------------\n",
       "    5                  60                          20465                  2            2            2             4             4             4"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": []
     },
     "execution_count": 313,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "gbm_classifier.summary()"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python [conda env:test_fiona5]",
   "language": "python",
   "name": "conda-env-test_fiona5-py"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 2
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython2",
   "version": "2.7.12"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}