final-notebook.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/laksh/ml/local/lib/python2.7/site-packages/sklearn/ensemble/weight_boosting.py:29: DeprecationWarning: numpy.core.umath_tests is an internal NumPy module and should not be imported. It will be removed in a future NumPy release.\n",
      "  from numpy.core.umath_tests import inner1d\n",
      "/home/laksh/ml/local/lib/python2.7/site-packages/sklearn/cross_validation.py:41: DeprecationWarning: This module was deprecated in version 0.18 in favor of the model_selection module into which all the refactored classes and functions are moved. Also note that the interface of the new CV iterators are different from that of this module. This module will be removed in 0.20.\n",
      "  \"This module will be removed in 0.20.\", DeprecationWarning)\n"
     ]
    }
   ],
   "source": [
    "# Importing all necessary modules\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "import random\n",
    "\n",
    "from sklearn.linear_model import LinearRegression as LR\n",
    "from sklearn.ensemble import RandomForestRegressor as RFR\n",
    "from sklearn.ensemble import GradientBoostingRegressor as GBR\n",
    "from sklearn.preprocessing import Imputer, LabelEncoder\n",
    "from sklearn.cross_validation import train_test_split, cross_val_predict, cross_val_score\n",
    "from sklearn.metrics import mean_squared_error\n",
    "from math import sqrt\n",
    "import xgboost as xgb\n",
    "\n",
    "%matplotlib inline\n",
    "sns.set()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Reading the data with custom data separator ';' \n",
    "df = pd.read_csv('./dataset.csv', sep=';')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style>\n",
       "    .dataframe thead tr:only-child th {\n",
       "        text-align: right;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: left;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>row_num</th>\n",
       "      <th>locale</th>\n",
       "      <th>day_of_week</th>\n",
       "      <th>hour_of_day</th>\n",
       "      <th>agent_id</th>\n",
       "      <th>entry_page</th>\n",
       "      <th>path_id_set</th>\n",
       "      <th>traffic_type</th>\n",
       "      <th>session_durantion</th>\n",
       "      <th>hits</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>988681</td>\n",
       "      <td>L6</td>\n",
       "      <td>Monday</td>\n",
       "      <td>17</td>\n",
       "      <td>1</td>\n",
       "      <td>2111</td>\n",
       "      <td>31672;0</td>\n",
       "      <td>6</td>\n",
       "      <td>7037</td>\n",
       "      <td>\\N</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>988680</td>\n",
       "      <td>L2</td>\n",
       "      <td>Thursday</td>\n",
       "      <td>22</td>\n",
       "      <td>10</td>\n",
       "      <td>2113</td>\n",
       "      <td>31965;0</td>\n",
       "      <td>2</td>\n",
       "      <td>49</td>\n",
       "      <td>14</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>988679</td>\n",
       "      <td>L4</td>\n",
       "      <td>Saturday</td>\n",
       "      <td>21</td>\n",
       "      <td>2</td>\n",
       "      <td>2100</td>\n",
       "      <td>0;78464</td>\n",
       "      <td>1</td>\n",
       "      <td>1892</td>\n",
       "      <td>14</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>988678</td>\n",
       "      <td>L3</td>\n",
       "      <td>Saturday</td>\n",
       "      <td>19</td>\n",
       "      <td>8</td>\n",
       "      <td>2113</td>\n",
       "      <td>51462</td>\n",
       "      <td>6</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>988677</td>\n",
       "      <td>L2</td>\n",
       "      <td>Tuesday</td>\n",
       "      <td>6</td>\n",
       "      <td>10</td>\n",
       "      <td>2116</td>\n",
       "      <td>31931;0</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   row_num locale day_of_week  hour_of_day  agent_id  entry_page path_id_set  \\\n",
       "0   988681     L6      Monday           17         1        2111     31672;0   \n",
       "1   988680     L2    Thursday           22        10        2113     31965;0   \n",
       "2   988679     L4    Saturday           21         2        2100     0;78464   \n",
       "3   988678     L3    Saturday           19         8        2113       51462   \n",
       "4   988677     L2     Tuesday            6        10        2116     31931;0   \n",
       "\n",
       "   traffic_type session_durantion hits  \n",
       "0             6              7037   \\N  \n",
       "1             2                49   14  \n",
       "2             1              1892   14  \n",
       "3             6                 0    1  \n",
       "4             1                 2    3  "
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Top 5 rows of the dataset\n",
    "df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style>\n",
       "    .dataframe thead tr:only-child th {\n",
       "        text-align: right;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: left;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>row_num</th>\n",
       "      <th>locale</th>\n",
       "      <th>day_of_week</th>\n",
       "      <th>hour_of_day</th>\n",
       "      <th>agent_id</th>\n",
       "      <th>entry_page</th>\n",
       "      <th>path_id_set</th>\n",
       "      <th>traffic_type</th>\n",
       "      <th>session_durantion</th>\n",
       "      <th>hits</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>988676</th>\n",
       "      <td>5</td>\n",
       "      <td>L2</td>\n",
       "      <td>Tuesday</td>\n",
       "      <td>18</td>\n",
       "      <td>2</td>\n",
       "      <td>2113</td>\n",
       "      <td>44232;0</td>\n",
       "      <td>2</td>\n",
       "      <td>5</td>\n",
       "      <td>\\N</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>988677</th>\n",
       "      <td>4</td>\n",
       "      <td>L6</td>\n",
       "      <td>Tuesday</td>\n",
       "      <td>9</td>\n",
       "      <td>6</td>\n",
       "      <td>2113</td>\n",
       "      <td>38715;0;40057;31657</td>\n",
       "      <td>2</td>\n",
       "      <td>35</td>\n",
       "      <td>\\N</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>988678</th>\n",
       "      <td>3</td>\n",
       "      <td>L2</td>\n",
       "      <td>Sunday</td>\n",
       "      <td>17</td>\n",
       "      <td>8</td>\n",
       "      <td>2114</td>\n",
       "      <td>15466;0</td>\n",
       "      <td>6</td>\n",
       "      <td>1011</td>\n",
       "      <td>\\N</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>988679</th>\n",
       "      <td>2</td>\n",
       "      <td>L3</td>\n",
       "      <td>Monday</td>\n",
       "      <td>17</td>\n",
       "      <td>13</td>\n",
       "      <td>2113</td>\n",
       "      <td>531228;0</td>\n",
       "      <td>2</td>\n",
       "      <td>132</td>\n",
       "      <td>\\N</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>988680</th>\n",
       "      <td>1</td>\n",
       "      <td>L4</td>\n",
       "      <td>Thursday</td>\n",
       "      <td>15</td>\n",
       "      <td>1</td>\n",
       "      <td>2116</td>\n",
       "      <td>76673;0</td>\n",
       "      <td>2</td>\n",
       "      <td>4544</td>\n",
       "      <td>18</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "        row_num locale day_of_week  hour_of_day  agent_id  entry_page  \\\n",
       "988676        5     L2     Tuesday           18         2        2113   \n",
       "988677        4     L6     Tuesday            9         6        2113   \n",
       "988678        3     L2      Sunday           17         8        2114   \n",
       "988679        2     L3      Monday           17        13        2113   \n",
       "988680        1     L4    Thursday           15         1        2116   \n",
       "\n",
       "                path_id_set  traffic_type session_durantion hits  \n",
       "988676              44232;0             2                 5   \\N  \n",
       "988677  38715;0;40057;31657             2                35   \\N  \n",
       "988678              15466;0             6              1011   \\N  \n",
       "988679             531228;0             2               132   \\N  \n",
       "988680              76673;0             2              4544   18  "
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Last 5 rows of the dataset\n",
    "df.tail()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style>\n",
       "    .dataframe thead tr:only-child th {\n",
       "        text-align: right;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: left;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>row_num</th>\n",
       "      <th>locale</th>\n",
       "      <th>day_of_week</th>\n",
       "      <th>hour_of_day</th>\n",
       "      <th>agent_id</th>\n",
       "      <th>entry_page</th>\n",
       "      <th>path_id_set</th>\n",
       "      <th>traffic_type</th>\n",
       "      <th>session_durantion</th>\n",
       "      <th>hits</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>542053</th>\n",
       "      <td>446628</td>\n",
       "      <td>L4</td>\n",
       "      <td>Monday</td>\n",
       "      <td>17</td>\n",
       "      <td>10</td>\n",
       "      <td>2116</td>\n",
       "      <td>76497;0</td>\n",
       "      <td>1</td>\n",
       "      <td>10</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>804561</th>\n",
       "      <td>184120</td>\n",
       "      <td>L4</td>\n",
       "      <td>Tuesday</td>\n",
       "      <td>11</td>\n",
       "      <td>9</td>\n",
       "      <td>2113</td>\n",
       "      <td>78506;0</td>\n",
       "      <td>4</td>\n",
       "      <td>224</td>\n",
       "      <td>11</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>799719</th>\n",
       "      <td>188962</td>\n",
       "      <td>L3</td>\n",
       "      <td>Wednesday</td>\n",
       "      <td>7</td>\n",
       "      <td>10</td>\n",
       "      <td>2116</td>\n",
       "      <td>34759;0</td>\n",
       "      <td>4</td>\n",
       "      <td>16</td>\n",
       "      <td>5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>238338</th>\n",
       "      <td>750343</td>\n",
       "      <td>L2</td>\n",
       "      <td>Wednesday</td>\n",
       "      <td>7</td>\n",
       "      <td>2</td>\n",
       "      <td>2111</td>\n",
       "      <td>31672;0</td>\n",
       "      <td>6</td>\n",
       "      <td>6</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>943273</th>\n",
       "      <td>45408</td>\n",
       "      <td>L3</td>\n",
       "      <td>Wednesday</td>\n",
       "      <td>23</td>\n",
       "      <td>9</td>\n",
       "      <td>2111</td>\n",
       "      <td>34227;0</td>\n",
       "      <td>6</td>\n",
       "      <td>32</td>\n",
       "      <td>5</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "        row_num locale day_of_week  hour_of_day  agent_id  entry_page  \\\n",
       "542053   446628     L4      Monday           17        10        2116   \n",
       "804561   184120     L4     Tuesday           11         9        2113   \n",
       "799719   188962     L3   Wednesday            7        10        2116   \n",
       "238338   750343     L2   Wednesday            7         2        2111   \n",
       "943273    45408     L3   Wednesday           23         9        2111   \n",
       "\n",
       "       path_id_set  traffic_type session_durantion hits  \n",
       "542053     76497;0             1                10    3  \n",
       "804561     78506;0             4               224   11  \n",
       "799719     34759;0             4                16    5  \n",
       "238338     31672;0             6                 6    3  \n",
       "943273     34227;0             6                32    5  "
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Random 5 rows of the dataset\n",
    "df.sample(5)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "By printing top and bottom rows of our dataset we can see that our dataset is in wrong order, let's fix that first!"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style>\n",
       "    .dataframe thead tr:only-child th {\n",
       "        text-align: right;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: left;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>row_num</th>\n",
       "      <th>locale</th>\n",
       "      <th>day_of_week</th>\n",
       "      <th>hour_of_day</th>\n",
       "      <th>agent_id</th>\n",
       "      <th>entry_page</th>\n",
       "      <th>path_id_set</th>\n",
       "      <th>traffic_type</th>\n",
       "      <th>session_durantion</th>\n",
       "      <th>hits</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1</td>\n",
       "      <td>L4</td>\n",
       "      <td>Thursday</td>\n",
       "      <td>15</td>\n",
       "      <td>1</td>\n",
       "      <td>2116</td>\n",
       "      <td>76673;0</td>\n",
       "      <td>2</td>\n",
       "      <td>4544</td>\n",
       "      <td>18</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2</td>\n",
       "      <td>L3</td>\n",
       "      <td>Monday</td>\n",
       "      <td>17</td>\n",
       "      <td>13</td>\n",
       "      <td>2113</td>\n",
       "      <td>531228;0</td>\n",
       "      <td>2</td>\n",
       "      <td>132</td>\n",
       "      <td>\\N</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>3</td>\n",
       "      <td>L2</td>\n",
       "      <td>Sunday</td>\n",
       "      <td>17</td>\n",
       "      <td>8</td>\n",
       "      <td>2114</td>\n",
       "      <td>15466;0</td>\n",
       "      <td>6</td>\n",
       "      <td>1011</td>\n",
       "      <td>\\N</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>4</td>\n",
       "      <td>L6</td>\n",
       "      <td>Tuesday</td>\n",
       "      <td>9</td>\n",
       "      <td>6</td>\n",
       "      <td>2113</td>\n",
       "      <td>38715;0;40057;31657</td>\n",
       "      <td>2</td>\n",
       "      <td>35</td>\n",
       "      <td>\\N</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>5</td>\n",
       "      <td>L2</td>\n",
       "      <td>Tuesday</td>\n",
       "      <td>18</td>\n",
       "      <td>2</td>\n",
       "      <td>2113</td>\n",
       "      <td>44232;0</td>\n",
       "      <td>2</td>\n",
       "      <td>5</td>\n",
       "      <td>\\N</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   row_num locale day_of_week  hour_of_day  agent_id  entry_page  \\\n",
       "0        1     L4    Thursday           15         1        2116   \n",
       "1        2     L3      Monday           17        13        2113   \n",
       "2        3     L2      Sunday           17         8        2114   \n",
       "3        4     L6     Tuesday            9         6        2113   \n",
       "4        5     L2     Tuesday           18         2        2113   \n",
       "\n",
       "           path_id_set  traffic_type session_durantion hits  \n",
       "0              76673;0             2              4544   18  \n",
       "1             531228;0             2               132   \\N  \n",
       "2              15466;0             6              1011   \\N  \n",
       "3  38715;0;40057;31657             2                35   \\N  \n",
       "4              44232;0             2                 5   \\N  "
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Flipping the whole dataset upside down and fixing the order\n",
    "df = df.iloc[::-1].reset_index(drop=True)\n",
    "df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style>\n",
       "    .dataframe thead tr:only-child th {\n",
       "        text-align: right;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: left;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>row_num</th>\n",
       "      <th>hour_of_day</th>\n",
       "      <th>agent_id</th>\n",
       "      <th>entry_page</th>\n",
       "      <th>traffic_type</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>count</th>\n",
       "      <td>988681.000000</td>\n",
       "      <td>988681.000000</td>\n",
       "      <td>988681.000000</td>\n",
       "      <td>988681.000000</td>\n",
       "      <td>988681.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mean</th>\n",
       "      <td>494341.000000</td>\n",
       "      <td>13.203988</td>\n",
       "      <td>7.350779</td>\n",
       "      <td>2253.265114</td>\n",
       "      <td>2.773766</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>std</th>\n",
       "      <td>285407.765084</td>\n",
       "      <td>6.718056</td>\n",
       "      <td>3.869630</td>\n",
       "      <td>775.083390</td>\n",
       "      <td>1.920976</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>min</th>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>2100.000000</td>\n",
       "      <td>1.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25%</th>\n",
       "      <td>247171.000000</td>\n",
       "      <td>8.000000</td>\n",
       "      <td>6.000000</td>\n",
       "      <td>2111.000000</td>\n",
       "      <td>1.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>50%</th>\n",
       "      <td>494341.000000</td>\n",
       "      <td>14.000000</td>\n",
       "      <td>9.000000</td>\n",
       "      <td>2113.000000</td>\n",
       "      <td>2.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>75%</th>\n",
       "      <td>741511.000000</td>\n",
       "      <td>19.000000</td>\n",
       "      <td>10.000000</td>\n",
       "      <td>2116.000000</td>\n",
       "      <td>4.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>max</th>\n",
       "      <td>988681.000000</td>\n",
       "      <td>23.000000</td>\n",
       "      <td>15.000000</td>\n",
       "      <td>8101.000000</td>\n",
       "      <td>10.000000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "             row_num    hour_of_day       agent_id     entry_page  \\\n",
       "count  988681.000000  988681.000000  988681.000000  988681.000000   \n",
       "mean   494341.000000      13.203988       7.350779    2253.265114   \n",
       "std    285407.765084       6.718056       3.869630     775.083390   \n",
       "min         1.000000       0.000000       0.000000    2100.000000   \n",
       "25%    247171.000000       8.000000       6.000000    2111.000000   \n",
       "50%    494341.000000      14.000000       9.000000    2113.000000   \n",
       "75%    741511.000000      19.000000      10.000000    2116.000000   \n",
       "max    988681.000000      23.000000      15.000000    8101.000000   \n",
       "\n",
       "        traffic_type  \n",
       "count  988681.000000  \n",
       "mean        2.773766  \n",
       "std         1.920976  \n",
       "min         1.000000  \n",
       "25%         1.000000  \n",
       "50%         2.000000  \n",
       "75%         4.000000  \n",
       "max        10.000000  "
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Let's look at the numeric columns of the dataset and some of it's statistics\n",
    "df.describe()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "RangeIndex: 988681 entries, 0 to 988680\n",
      "Data columns (total 10 columns):\n",
      "row_num              988681 non-null int64\n",
      "locale               988681 non-null object\n",
      "day_of_week          988681 non-null object\n",
      "hour_of_day          988681 non-null int64\n",
      "agent_id             988681 non-null int64\n",
      "entry_page           988681 non-null int64\n",
      "path_id_set          983792 non-null object\n",
      "traffic_type         988681 non-null int64\n",
      "session_durantion    988681 non-null object\n",
      "hits                 988681 non-null object\n",
      "dtypes: int64(5), object(5)\n",
      "memory usage: 75.4+ MB\n"
     ]
    }
   ],
   "source": [
    "# Let's look at the some info regarding the dataset such count of values in each columns and it's type\n",
    "df.info()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array(['18', '\\\\N', '2', '37', '69', '3', '25', '33', '1', '11', '5',\n",
       "       '22', '70', '6', '4', '27', '8', '12', '20', '13', '126', '7',\n",
       "       '17', '56', '21', '10', '279', '28', '31', '16', '236', '40', '15',\n",
       "       '14', '87', '30', '9', '45', '51', '32', '46', '42', '23', '52',\n",
       "       '60', '24', '63', '39', '19', '107', '80', '76', '26', '53', '41',\n",
       "       '58', '35', '44', '143', '77', '43', '57', '36', '29', '248', '61',\n",
       "       '138', '148', '363', '48', '78', '47', '103', '86', '131', '50',\n",
       "       '109', '132', '64', '99', '34', '95', '49', '93', '62', '67',\n",
       "       '105', '120', '88', '59', '38', '114', '96', '75', '212', '135',\n",
       "       '169', '89', '54', '179', '85', '92', '112', '118', '128', '375',\n",
       "       '359', '55', '382', '82', '916', '111', '130', '174', '71', '98',\n",
       "       '83', '73', '137', '108', '159', '81', '317', '218', '104', '66',\n",
       "       '127', '150', '225', '115', '121', '175', '182', '74', '117', '68',\n",
       "       '160', '504', '101', '152', '141', '116', '217', '100', '149',\n",
       "       '226', '129', '166', '187', '157', '393', '134', '551', '94',\n",
       "       '106', '72', '84', '246', '221', '65', '181', '205', '90', '119',\n",
       "       '102', '315', '324', '113', '110', '337', '147', '156', '255',\n",
       "       '196', '222', '97', '142', '233', '200', '145', '262', '197',\n",
       "       '256', '291', '155', '191', '140', '91', '146', '282', '79', '192',\n",
       "       '161', '388', '162', '342', '194', '177', '243', '193', '178',\n",
       "       '343', '517', '186', '151', '153', '154', '123', '561', '265',\n",
       "       '124', '254', '208', '287', '345', '176', '144', '185', '171',\n",
       "       '136', '167', '280', '125', '189', '496', '168', '173', '237',\n",
       "       '184', '220', '158', '384', '139', '385', '289', '586', '632',\n",
       "       '355', '252', '340', '229', '413', '207', '122', '349', '172',\n",
       "       '249', '202', '266', '170', '391', '224', '234', '459', '239',\n",
       "       '163', '203', '164', '325', '268', '247', '251', '133', '447',\n",
       "       '165', '516', '210', '190', '528', '240', '264', '214', '668',\n",
       "       '339', '481', '231', '323', '566', '195', '211', '257', '223',\n",
       "       '426', '297', '671', '244', '302', '281', '310', '253', '276',\n",
       "       '608', '245', '180', '204', '666', '799', '284', '428', '183',\n",
       "       '794', '260', '332', '331', '667', '272', '589', '974', '841',\n",
       "       '206', '199', '232', '330', '947', '372', '414', '401', '461',\n",
       "       '364', '677', '294', '1300', '219', '271', '242', '607', '417',\n",
       "       '705', '653', '259', '300', '350', '209', '263', '215', '228',\n",
       "       '213', '261', '552', '454', '575', '451', '335', '420', '316',\n",
       "       '198', '293', '605', '321', '449', '238', '480', '274', '273',\n",
       "       '290', '513', '432', '369', '267', '405', '557', '767', '868',\n",
       "       '201', '277', '314', '489', '308', '365', '306', '398', '303',\n",
       "       '399', '286', '352', '701', '295', '216', '275', '587', '304',\n",
       "       '1714', '471', '188', '400', '351', '250', '499', '445', '328',\n",
       "       '312', '483', '288', '442', '322', '379', '374', '469', '745',\n",
       "       '410', '386', '404', '320', '665', '230', '407', '425', '509',\n",
       "       '338', '519', '4174', '1069', '563', '347', '472', '415', '299',\n",
       "       '348', '269', '473', '422', '395', '514', '333', '487', '412',\n",
       "       '357', '380', '319', '381', '227', '629', '495', '356', '296',\n",
       "       '1118', '278', '485', '2072', '423', '792', '453', '511', '518',\n",
       "       '462', '733', '698', '879', '530', '258', '1525', '532', '235',\n",
       "       '446', '995', '411', '576', '285', '241', '624', '361', '311',\n",
       "       '309', '370', '307', '564', '468', '512', '717', '807', '1077',\n",
       "       '394', '362', '437', '378', '283', '301', '433', '358', '270',\n",
       "       '616', '565', '593', '558', '549', '488', '617', '817', '2081',\n",
       "       '341', '292', '610', '334', '662', '353', '367', '505', '1179',\n",
       "       '539', '693', '326', '354', '727', '585', '546', '440', '371',\n",
       "       '929', '763', '390', '329', '344', '475', '681', '560', '327',\n",
       "       '376', '820', '313', '965', '1276', '318', '494', '534', '515',\n",
       "       '373', '305', '403', '434', '436', '491', '783', '630', '679',\n",
       "       '418', '336', '798', '842', '298', '1259', '731', '435', '377',\n",
       "       '476', '540', '1027', '814', '1093', '571', '548', '456', '366',\n",
       "       '416', '368', '718', '438', '932', '796', '578', '690', '419',\n",
       "       '606', '568', '2320', '1147', '1083', '760', '1975', '527', '559',\n",
       "       '750', '697', '431', '421', '346', '567', '765', '658', '521',\n",
       "       '580', '402', '465', '562', '1051', '574', '525', '582', '866',\n",
       "       '547', '1008', '464', '470', '768', '809', '824', '780', '441',\n",
       "       '1606', '687', '625', '424', '533', '537', '674', '392', '588',\n",
       "       '409', '484', '634', '544', '570', '408', '694', '429', '531',\n",
       "       '784', '785', '477', '899', '1450', '520', '1214', '2603', '360',\n",
       "       '389', '1071', '493', '678', '804', '654', '479', '894', '452',\n",
       "       '940', '660', '650', '603', '715', '1301', '858', '439', '396',\n",
       "       '406', '797', '502', '1068', '383', '458', '920', '826', '597',\n",
       "       '801', '702', '670', '448', '450', '1000', '599', '467', '510',\n",
       "       '503', '661', '626', '600', '743', '735', '639', '2108', '852',\n",
       "       '529', '726', '655', '501', '1923', '827', '466', '536', '1878',\n",
       "       '675', '708', '1142', '543', '1079', '1244', '594', '474', '815',\n",
       "       '1108', '387', '443', '637', '989', '1303', '497', '444', '623',\n",
       "       '427', '649', '648', '1721', '782', '1712', '704', '1445', '1449',\n",
       "       '1173', '1749', '791', '478', '742', '591', '1030', '620', '683',\n",
       "       '984', '849', '595', '747', '751', '611', '645', '754', '757',\n",
       "       '1213', '788', '983', '507', '1203', '758', '455', '1811', '1345',\n",
       "       '640', '524', '770', '925', '573', '602', '721', '598', '554',\n",
       "       '1557', '728', '1451', '1054', '550', '1009', '615', '734', '688',\n",
       "       '613', '769', '822', '738', '2049', '1523', '596', '1248', '847',\n",
       "       '790', '583', '921', '689', '541', '872', '482', '633', '579',\n",
       "       '902', '719', '2521', '891', '1085', '778', '641', '2384', '802',\n",
       "       '887', '714', '930', '556', '506', '1037', '460', '592', '997',\n",
       "       '1343', '1275', '490', '584', '553', '636', '1554', '680', '522',\n",
       "       '669', '691', '604', '569', '912', '590', '969', '622', '896',\n",
       "       '430', '572', '1073', '535', '823', '621', '498', '1036', '713'],\n",
       "      dtype=object)"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df['hits'].unique()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "According to the Pandas above **.info()** function each and every column is non-null but when we see the unique values of **hits** and **session duration** column we see that it contains a pretty unique value which is **\\\\\\N** and is being used as a placeholder for empty/nan value. We will deal it with later."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Let's do some visualizations and see if there are any patterns."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x7fc7bb285110>"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAe8AAAFLCAYAAAD/BAQ1AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nAAAYhvIGAMAwlDcAAIb5f+G0FyVPPqJBAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fc7bb673550>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "df['hour_of_day'].value_counts().sort_index().plot(kind='bar')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The above bar graph shows that people least visit the website around **2-10am** but peaks between **7-9pm** "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x7fc7bb159d90>"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAfUAAAGDCAYAAAAyM4nNAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nLiIiYhL/D6+PFFyz54QUAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fc7bb106a50>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "weekdays = ['Monday', 'Tuesday', 'Wednesday', 'Thursday', 'Friday', 'Saturday', 'Sunday']\n",
    "mapping = {day: i for i, day in enumerate(weekdays)}\n",
    "key = df['day_of_week'].map(mapping)\n",
    "df.iloc[key.argsort()].groupby('day_of_week', sort=False).size().plot(kind='bar')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x7fc7bafc9810>"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAfUAAAFLCAYAAADPiBUUAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nABiCUAcAwBD/D6ERITqW31OMAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fc7bb1dc790>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "df['locale'].value_counts().sort_index().plot(kind='bar')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x7fc7baf1d290>"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAfUAAAFNCAYAAAAZ0fYJAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fc7bb2c1090>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "df['agent_id'].value_counts().sort_index().plot(kind='bar')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Total number of values in 'hits' column: 988681 (100%)\n",
      "Training Values:                         619235 (62.63%)\n",
      "Testing Values:                          369446 (37.37%)\n"
     ]
    }
   ],
   "source": [
    "# Let's see the number of rows for Training and Testing dataframe\n",
    "total_count = df['hits'].count()\n",
    "missing = df['hits'].value_counts()['\\N']\n",
    "useful = total_count - missing\n",
    "\n",
    "print \"Total number of values in 'hits' column: %d (%d%%)\" % (total_count, (total_count/ total_count) * 100)\n",
    "print \"Training Values:                         %d (%.2f%%)\" % (useful, (useful/float(total_count))*100)\n",
    "print \"Testing Values:                          %d (%.2f%%)\" % (missing, (missing/float(total_count))*100)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Create a new column which represent if row is Trainable or not\n",
    "dataset = df.copy()\n",
    "dataset['is_train'] = dataset['hits'] != '\\\\N'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Replace \\N to np.nan so that modules can handle these values\n",
    "dataset = dataset.replace('\\\\N', np.nan)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "row_num                   0\n",
       "locale                    0\n",
       "day_of_week               0\n",
       "hour_of_day               0\n",
       "agent_id                  0\n",
       "entry_page                0\n",
       "path_id_set            4889\n",
       "traffic_type              0\n",
       "session_durantion       668\n",
       "hits                 369446\n",
       "is_train                  0\n",
       "dtype: int64"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Counting how many nan values are there in the dataset in each columns\n",
    "dataset.isnull().sum()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Generating a new column based on session_duration and path_id_set column, this\n",
    "# extra information can help our trees divide the dataset better and learn better\n",
    "\n",
    "dataset['session_isnull'] = dataset['session_durantion'] == np.nan\n",
    "dataset['path_isnull'] = dataset['path_id_set'] == np.nan"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# In the top **info()** function we saw that our few columns such as \n",
    "# session_duration or hits are not int type so let's convert them\n",
    "\n",
    "dataset['session_durantion'] = dataset['session_durantion'].apply(pd.to_numeric, errors='coerce')\n",
    "dataset['hits'] = dataset['hits'].apply(pd.to_numeric, errors='coerce')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Let's replace nan values with some arbitary big negative \n",
    "# value which will be handled by our tree models\n",
    "\n",
    "cols = list(set(dataset.columns) - set(['hits']))\n",
    "dataset[cols] = dataset[cols].fillna(-999)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Converting locale and traffic type column to frequency encoding\n",
    "# useful than other encoding methods such as LabelEncoding as\n",
    "# it keeps the information such as the frequency a particular category\n",
    "# is present in the dataset\n",
    "\n",
    "freq_encoding = dataset.groupby('locale').size()\n",
    "freq_encoding = freq_encoding / len(dataset)\n",
    "dataset['locale_enc'] = dataset.locale.map(freq_encoding)\n",
    "\n",
    "freq_encoding = dataset.groupby('traffic_type').size()\n",
    "freq_encoding = freq_encoding / len(dataset)\n",
    "dataset['traffic_enc'] = dataset.traffic_type.map(freq_encoding)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Normaly some time, day of the week or particular hour is not useful because\n",
    "# time is a continues cyclic feature which is hard to represent just in some \n",
    "# numbers or LabelEncoder. I used sin and cos function to represent the time \n",
    "# and week of the day in a cyclic and repetitive manner. This code first encodes\n",
    "# weekdays and time as Mon 00:00 as 0.0 and Sun 23:59 as 0.9999 then using\n",
    "# sin and cos function to convert that range from 0-1 to sin/ cos function range\n",
    "# which makes it cyclic pattern.\n",
    "\n",
    "label_dict = {'Monday': float(0)/7, 'Tuesday': float(1)/7, 'Wednesday': float(2)/7, 'Thursday': float(3)/7,\n",
    "              'Friday': float(4)/7, 'Saturday': float(5)/7, 'Sunday': float(6)/7}\n",
    "\n",
    "dataset['week_enc'] = dataset['day_of_week']\n",
    "for day in dataset['day_of_week'].unique():\n",
    "    dataset['week_enc'] = dataset['week_enc'].replace(day, label_dict[day])\n",
    "\n",
    "hours_in_day = 24\n",
    "dataset['combined_sin'] = np.sin(dataset.week_enc*(dataset.hour_of_day*(2.*np.pi/hours_in_day)))\n",
    "dataset['combined_cos'] = np.cos(dataset.week_enc*(dataset.hour_of_day*(2.*np.pi/hours_in_day)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAWcAAAFYCAYAAACPlD31AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nRK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fc7bb12cf10>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# This graph shows the cyclic nature of day of the week and hour of the day\n",
    "dataset.sample(1000).plot.scatter('combined_sin', 'combined_cos').set_aspect('equal')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Counted total number of pages visit in a session and made a new feature\n",
    "# count_pages which is more useful than encoding paths in some way.\n",
    "\n",
    "dataset.loc[::, 'count_page'] = dataset.loc[::, 'path_id_set'].apply(lambda x: len(x.split(';')) if isinstance(x, basestring) else 0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Generating a new feature using entry page and session duration by multiplying\n",
    "# This will give our model easy to find some pattern between 2 feature relation\n",
    "# Entry page and Session Duration\n",
    "\n",
    "entry_session = dataset['entry_page'] * dataset['session_durantion']\n",
    "dataset.insert(loc=1, column='entry_session', value=entry_session)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Distributing dataset between training and testing (submit)\n",
    "\n",
    "train = dataset[~np.isnan(dataset['hits'])]\n",
    "test = dataset[np.isnan(dataset['hits'])]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Dropping non numeric columns which are not being used for training purposes\n",
    "\n",
    "X = train.drop(['hits', 'day_of_week', 'locale', 'path_id_set', 'is_train'], axis=1)\n",
    "y = pd.DataFrame(train['hits'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Shuffling and splitting dataset for validation while training the model\n",
    "\n",
    "X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.33, random_state=42)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "# Coverting the dataset to matrix form for XGBoost Model\n",
    "\n",
    "dtrain = xgb.DMatrix(X_train, label=y_train)\n",
    "dtest = xgb.DMatrix(X_test, label=y_test)\n",
    "\n",
    "# Some default params\n",
    "\n",
    "params = {\n",
    "    'max_depth':6,\n",
    "    'min_child_weight': 1,\n",
    "    'eta':.01,\n",
    "    'subsample': 1,\n",
    "    'colsample_bytree': 1,\n",
    "    'silent': 1,\n",
    "    'objective':'gpu:reg:linear',\n",
    "}\n",
    "\n",
    "params['eval_metric'] = \"rmse\"\n",
    "num_boost_round = 9999"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Searching different values and best values for multiple params\n",
    "\n",
    "#################### MAX DEPTH and MIN CHILD WEIGHT##########################\n",
    "gridsearch_params = [\n",
    "    (max_depth, min_child_weight)\n",
    "    for max_depth in range(3, 12)\n",
    "    for min_child_weight in range(1, 8)\n",
    "]\n",
    "\n",
    "min_rmse = float(\"Inf\")\n",
    "best_params = None\n",
    "for max_depth, min_child_weight in gridsearch_params:\n",
    "    print(\"CV with max_depth={}, min_child_weight={}\".format(\n",
    "                             max_depth,\n",
    "                             min_child_weight))\n",
    "\n",
    "    # Update our parameters\n",
    "    params['max_depth'] = max_depth\n",
    "    params['min_child_weight'] = min_child_weight\n",
    "\n",
    "    # Run CV\n",
    "    cv_results = xgb.cv(\n",
    "        params,\n",
    "        dtrain,\n",
    "        num_boost_round=num_boost_round,\n",
    "        seed=42,\n",
    "        nfold=5,\n",
    "        metrics={'rmse'},\n",
    "        early_stopping_rounds=20\n",
    "    )\n",
    "\n",
    "    # Update best rmse\n",
    "    mean_rmse = cv_results['test-rmse-mean'].min()\n",
    "    boost_rounds = cv_results['test-rmse-mean'].argmin()\n",
    "    print(\"\\trmse {} for {} rounds\".format(mean_rmse, boost_rounds))\n",
    "    if mean_rmse < min_rmse:\n",
    "        min_rmse = mean_rmse\n",
    "        best_params = (max_depth,min_child_weight)\n",
    "\n",
    "print(\"Best params: {}, {}, rmse: {}\".format(best_params[0], best_params[1], min_rmse))\n",
    "params['max_depth'] = best_params[0]\n",
    "params['min_child_weight'] = best_params[1]\n",
    "\n",
    "\n",
    "######################SUB SAMPLE AND COLS SAMPLE#######################################\n",
    "\n",
    "gridsearch_params = [\n",
    "    (subsample, colsample)\n",
    "    for subsample in [i/10. for i in range(7,11)]\n",
    "    for colsample in [i/10. for i in range(7,11)]\n",
    "]\n",
    "\n",
    "min_rmse = float(\"Inf\")\n",
    "best_params = None\n",
    "\n",
    "for subsample, colsample in reversed(gridsearch_params):\n",
    "    print(\"CV with subsample={}, colsample={}\".format(\n",
    "                             subsample,\n",
    "                             colsample))\n",
    "\n",
    "    params['subsample'] = subsample\n",
    "    params['colsample_bytree'] = colsample\n",
    "\n",
    "    cv_results = xgb.cv(\n",
    "        params,\n",
    "        dtrain,\n",
    "        num_boost_round=num_boost_round,\n",
    "        seed=42,\n",
    "        nfold=5,\n",
    "        metrics={'rmse'},\n",
    "        early_stopping_rounds=50\n",
    "    )\n",
    "\n",
    "    mean_rmse = cv_results['test-rmse-mean'].min()\n",
    "    boost_rounds = cv_results['test-rmse-mean'].argmin()\n",
    "    print(\"\\trmse {} for {} rounds\".format(mean_rmse, boost_rounds))\n",
    "    if mean_rmse < min_rmse:\n",
    "        min_rmse = mean_rmse\n",
    "        best_params = (subsample,colsample)\n",
    "\n",
    "print(\"Best params: {}, {}, rmse: {}\".format(best_params[0], best_params[1], min_rmse))\n",
    "params['subsample'] = best_params[0]\n",
    "params['colsample_bytree'] = best_params[0]\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Our best params: {'colsample_bytree': 1, 'silent': 1, 'eval_metric': 'rmse', 'min_child_weight': 1, 'subsample': 1, 'eta': 0.01, 'objective': 'gpu:reg:linear', 'max_depth': 6}\n",
      "[0]\tTest-rmse:42.5804\n",
      "Will train until Test-rmse hasn't improved in 20 rounds.\n",
      "[1]\tTest-rmse:42.3948\n",
      "[2]\tTest-rmse:42.2122\n",
      "[3]\tTest-rmse:42.0315\n",
      "[4]\tTest-rmse:41.8545\n",
      "[5]\tTest-rmse:41.6792\n",
      "[6]\tTest-rmse:41.5077\n",
      "[7]\tTest-rmse:41.3387\n",
      "[8]\tTest-rmse:41.1728\n",
      "[9]\tTest-rmse:41.0093\n",
      "[10]\tTest-rmse:40.8484\n",
      "[11]\tTest-rmse:40.6893\n",
      "[12]\tTest-rmse:40.5332\n",
      "[13]\tTest-rmse:40.3798\n",
      "[14]\tTest-rmse:40.229\n",
      "[15]\tTest-rmse:40.0803\n",
      "[16]\tTest-rmse:39.9335\n",
      "[17]\tTest-rmse:39.7901\n",
      "[18]\tTest-rmse:39.6477\n",
      "[19]\tTest-rmse:39.5086\n",
      "[20]\tTest-rmse:39.3705\n",
      "[21]\tTest-rmse:39.2355\n",
      "[22]\tTest-rmse:39.1033\n",
      "[23]\tTest-rmse:38.9734\n",
      "[24]\tTest-rmse:38.8449\n",
      "[25]\tTest-rmse:38.7185\n",
      "[26]\tTest-rmse:38.594\n",
      "[27]\tTest-rmse:38.4713\n",
      "[28]\tTest-rmse:38.3511\n",
      "[29]\tTest-rmse:38.2319\n",
      "[30]\tTest-rmse:38.1154\n",
      "[31]\tTest-rmse:38.0006\n",
      "[32]\tTest-rmse:37.8877\n",
      "[33]\tTest-rmse:37.7769\n",
      "[34]\tTest-rmse:37.6684\n",
      "[35]\tTest-rmse:37.5609\n",
      "[36]\tTest-rmse:37.4553\n",
      "[37]\tTest-rmse:37.352\n",
      "[38]\tTest-rmse:37.2503\n",
      "[39]\tTest-rmse:37.149\n",
      "[40]\tTest-rmse:37.0496\n",
      "[41]\tTest-rmse:36.9522\n",
      "[42]\tTest-rmse:36.8567\n",
      "[43]\tTest-rmse:36.7626\n",
      "[44]\tTest-rmse:36.6709\n",
      "[45]\tTest-rmse:36.5792\n",
      "[46]\tTest-rmse:36.4895\n",
      "[47]\tTest-rmse:36.4011\n",
      "[48]\tTest-rmse:36.3147\n",
      "[49]\tTest-rmse:36.2308\n",
      "[50]\tTest-rmse:36.147\n",
      "[51]\tTest-rmse:36.0648\n",
      "[52]\tTest-rmse:35.9838\n",
      "[53]\tTest-rmse:35.9046\n",
      "[54]\tTest-rmse:35.8267\n",
      "[55]\tTest-rmse:35.7497\n",
      "[56]\tTest-rmse:35.6746\n",
      "[57]\tTest-rmse:35.5998\n",
      "[58]\tTest-rmse:35.5271\n",
      "[59]\tTest-rmse:35.4559\n",
      "[60]\tTest-rmse:35.3852\n",
      "[61]\tTest-rmse:35.3153\n",
      "[62]\tTest-rmse:35.2481\n",
      "[63]\tTest-rmse:35.1818\n",
      "[64]\tTest-rmse:35.1156\n",
      "[65]\tTest-rmse:35.0506\n",
      "[66]\tTest-rmse:34.9875\n",
      "[67]\tTest-rmse:34.9258\n",
      "[68]\tTest-rmse:34.8642\n",
      "[69]\tTest-rmse:34.8033\n",
      "[70]\tTest-rmse:34.7444\n",
      "[71]\tTest-rmse:34.6869\n",
      "[72]\tTest-rmse:34.6284\n",
      "[73]\tTest-rmse:34.5726\n",
      "[74]\tTest-rmse:34.5166\n",
      "[75]\tTest-rmse:34.4619\n",
      "[76]\tTest-rmse:34.4094\n",
      "[77]\tTest-rmse:34.3558\n",
      "[78]\tTest-rmse:34.3038\n",
      "[79]\tTest-rmse:34.2522\n",
      "[80]\tTest-rmse:34.203\n",
      "[81]\tTest-rmse:34.1541\n",
      "[82]\tTest-rmse:34.1062\n",
      "[83]\tTest-rmse:34.0582\n",
      "[84]\tTest-rmse:34.0122\n",
      "[85]\tTest-rmse:33.9665\n",
      "[86]\tTest-rmse:33.921\n",
      "[87]\tTest-rmse:33.8777\n",
      "[88]\tTest-rmse:33.8348\n",
      "[89]\tTest-rmse:33.7923\n",
      "[90]\tTest-rmse:33.7515\n",
      "[91]\tTest-rmse:33.7108\n",
      "[92]\tTest-rmse:33.6716\n",
      "[93]\tTest-rmse:33.6331\n",
      "[94]\tTest-rmse:33.5941\n",
      "[95]\tTest-rmse:33.556\n",
      "[96]\tTest-rmse:33.5187\n",
      "[97]\tTest-rmse:33.4828\n",
      "[98]\tTest-rmse:33.4474\n",
      "[99]\tTest-rmse:33.4122\n",
      "[100]\tTest-rmse:33.3772\n",
      "[101]\tTest-rmse:33.3432\n",
      "[102]\tTest-rmse:33.3097\n",
      "[103]\tTest-rmse:33.2769\n",
      "[104]\tTest-rmse:33.2448\n",
      "[105]\tTest-rmse:33.2129\n",
      "[106]\tTest-rmse:33.1831\n",
      "[107]\tTest-rmse:33.1524\n",
      "[108]\tTest-rmse:33.1227\n",
      "[109]\tTest-rmse:33.0937\n",
      "[110]\tTest-rmse:33.0654\n",
      "[111]\tTest-rmse:33.0367\n",
      "[112]\tTest-rmse:33.0081\n",
      "[113]\tTest-rmse:32.9816\n",
      "[114]\tTest-rmse:32.9543\n",
      "[115]\tTest-rmse:32.9273\n",
      "[116]\tTest-rmse:32.9016\n",
      "[117]\tTest-rmse:32.876\n",
      "[118]\tTest-rmse:32.8512\n",
      "[119]\tTest-rmse:32.8262\n",
      "[120]\tTest-rmse:32.8028\n",
      "[121]\tTest-rmse:32.7791\n",
      "[122]\tTest-rmse:32.7565\n",
      "[123]\tTest-rmse:32.7343\n",
      "[124]\tTest-rmse:32.7124\n",
      "[125]\tTest-rmse:32.6909\n",
      "[126]\tTest-rmse:32.6693\n",
      "[127]\tTest-rmse:32.6486\n",
      "[128]\tTest-rmse:32.6283\n",
      "[129]\tTest-rmse:32.6077\n",
      "[130]\tTest-rmse:32.5878\n",
      "[131]\tTest-rmse:32.5686\n",
      "[132]\tTest-rmse:32.55\n",
      "[133]\tTest-rmse:32.5305\n",
      "[134]\tTest-rmse:32.5117\n",
      "[135]\tTest-rmse:32.4936\n",
      "[136]\tTest-rmse:32.4757\n",
      "[137]\tTest-rmse:32.4573\n",
      "[138]\tTest-rmse:32.4403\n",
      "[139]\tTest-rmse:32.4234\n",
      "[140]\tTest-rmse:32.4068\n",
      "[141]\tTest-rmse:32.3904\n",
      "[142]\tTest-rmse:32.3753\n",
      "[143]\tTest-rmse:32.3589\n",
      "[144]\tTest-rmse:32.3428\n",
      "[145]\tTest-rmse:32.3285\n",
      "[146]\tTest-rmse:32.3135\n",
      "[147]\tTest-rmse:32.2997\n",
      "[148]\tTest-rmse:32.2854\n",
      "[149]\tTest-rmse:32.2719\n",
      "[150]\tTest-rmse:32.2574\n",
      "[151]\tTest-rmse:32.2449\n",
      "[152]\tTest-rmse:32.2316\n",
      "[153]\tTest-rmse:32.2187\n",
      "[154]\tTest-rmse:32.2054\n",
      "[155]\tTest-rmse:32.1932\n",
      "[156]\tTest-rmse:32.1815\n",
      "[157]\tTest-rmse:32.1695\n",
      "[158]\tTest-rmse:32.1575\n",
      "[159]\tTest-rmse:32.1466\n",
      "[160]\tTest-rmse:32.1349\n",
      "[161]\tTest-rmse:32.1236\n",
      "[162]\tTest-rmse:32.1123\n",
      "[163]\tTest-rmse:32.1015\n",
      "[164]\tTest-rmse:32.0906\n",
      "[165]\tTest-rmse:32.0798\n",
      "[166]\tTest-rmse:32.0693\n",
      "[167]\tTest-rmse:32.0593\n",
      "[168]\tTest-rmse:32.049\n",
      "[169]\tTest-rmse:32.0384\n",
      "[170]\tTest-rmse:32.029\n",
      "[171]\tTest-rmse:32.0191\n",
      "[172]\tTest-rmse:32.0093\n",
      "[173]\tTest-rmse:31.9999\n",
      "[174]\tTest-rmse:31.991\n",
      "[175]\tTest-rmse:31.9809\n",
      "[176]\tTest-rmse:31.972\n",
      "[177]\tTest-rmse:31.9635\n",
      "[178]\tTest-rmse:31.9553\n",
      "[179]\tTest-rmse:31.9473\n",
      "[180]\tTest-rmse:31.9401\n",
      "[181]\tTest-rmse:31.9324\n",
      "[182]\tTest-rmse:31.9246\n",
      "[183]\tTest-rmse:31.9167\n",
      "[184]\tTest-rmse:31.9084\n",
      "[185]\tTest-rmse:31.9008\n",
      "[186]\tTest-rmse:31.8931\n",
      "[187]\tTest-rmse:31.8852\n",
      "[188]\tTest-rmse:31.8785\n",
      "[189]\tTest-rmse:31.8719\n",
      "[190]\tTest-rmse:31.8642\n",
      "[191]\tTest-rmse:31.8576\n",
      "[192]\tTest-rmse:31.8509\n",
      "[193]\tTest-rmse:31.8448\n",
      "[194]\tTest-rmse:31.8388\n",
      "[195]\tTest-rmse:31.8328\n",
      "[196]\tTest-rmse:31.8265\n",
      "[197]\tTest-rmse:31.8209\n",
      "[198]\tTest-rmse:31.8154\n",
      "[199]\tTest-rmse:31.8089\n",
      "[200]\tTest-rmse:31.803\n",
      "[201]\tTest-rmse:31.7981\n",
      "[202]\tTest-rmse:31.7923\n",
      "[203]\tTest-rmse:31.7865\n",
      "[204]\tTest-rmse:31.7815\n",
      "[205]\tTest-rmse:31.7763\n",
      "[206]\tTest-rmse:31.7711\n",
      "[207]\tTest-rmse:31.7659\n",
      "[208]\tTest-rmse:31.7607\n",
      "[209]\tTest-rmse:31.7557\n",
      "[210]\tTest-rmse:31.7507\n",
      "[211]\tTest-rmse:31.7463\n",
      "[212]\tTest-rmse:31.7416\n",
      "[213]\tTest-rmse:31.7363\n",
      "[214]\tTest-rmse:31.7319\n",
      "[215]\tTest-rmse:31.7275\n",
      "[216]\tTest-rmse:31.7235\n",
      "[217]\tTest-rmse:31.7188\n",
      "[218]\tTest-rmse:31.715\n",
      "[219]\tTest-rmse:31.7106\n",
      "[220]\tTest-rmse:31.7057\n",
      "[221]\tTest-rmse:31.702\n",
      "[222]\tTest-rmse:31.6975\n",
      "[223]\tTest-rmse:31.6939\n",
      "[224]\tTest-rmse:31.6903\n",
      "[225]\tTest-rmse:31.6869\n",
      "[226]\tTest-rmse:31.6834\n",
      "[227]\tTest-rmse:31.6794\n",
      "[228]\tTest-rmse:31.6761\n",
      "[229]\tTest-rmse:31.6726\n",
      "[230]\tTest-rmse:31.6688\n",
      "[231]\tTest-rmse:31.6657\n",
      "[232]\tTest-rmse:31.6625\n",
      "[233]\tTest-rmse:31.6589\n",
      "[234]\tTest-rmse:31.6559\n",
      "[235]\tTest-rmse:31.6528\n",
      "[236]\tTest-rmse:31.6497\n",
      "[237]\tTest-rmse:31.6467\n",
      "[238]\tTest-rmse:31.644\n",
      "[239]\tTest-rmse:31.64\n",
      "[240]\tTest-rmse:31.6371\n",
      "[241]\tTest-rmse:31.6348\n",
      "[242]\tTest-rmse:31.6322\n",
      "[243]\tTest-rmse:31.6284\n",
      "[244]\tTest-rmse:31.6256\n",
      "[245]\tTest-rmse:31.6232\n",
      "[246]\tTest-rmse:31.6202\n",
      "[247]\tTest-rmse:31.6175\n",
      "[248]\tTest-rmse:31.6152\n",
      "[249]\tTest-rmse:31.6125\n",
      "[250]\tTest-rmse:31.6102\n",
      "[251]\tTest-rmse:31.6077\n",
      "[252]\tTest-rmse:31.6055\n",
      "[253]\tTest-rmse:31.6027\n",
      "[254]\tTest-rmse:31.6003\n",
      "[255]\tTest-rmse:31.5979\n",
      "[256]\tTest-rmse:31.5951\n",
      "[257]\tTest-rmse:31.5929\n",
      "[258]\tTest-rmse:31.5907\n",
      "[259]\tTest-rmse:31.5882\n",
      "[260]\tTest-rmse:31.5862\n",
      "[261]\tTest-rmse:31.5838\n",
      "[262]\tTest-rmse:31.5817\n",
      "[263]\tTest-rmse:31.5795\n",
      "[264]\tTest-rmse:31.5774\n",
      "[265]\tTest-rmse:31.5752\n",
      "[266]\tTest-rmse:31.5733\n",
      "[267]\tTest-rmse:31.5715\n",
      "[268]\tTest-rmse:31.5695\n",
      "[269]\tTest-rmse:31.5677\n",
      "[270]\tTest-rmse:31.5657\n",
      "[271]\tTest-rmse:31.5636\n",
      "[272]\tTest-rmse:31.5622\n",
      "[273]\tTest-rmse:31.5605\n",
      "[274]\tTest-rmse:31.5589\n",
      "[275]\tTest-rmse:31.557\n",
      "[276]\tTest-rmse:31.5551\n",
      "[277]\tTest-rmse:31.5535\n",
      "[278]\tTest-rmse:31.5519\n",
      "[279]\tTest-rmse:31.5504\n",
      "[280]\tTest-rmse:31.5486\n",
      "[281]\tTest-rmse:31.5468\n",
      "[282]\tTest-rmse:31.5456\n",
      "[283]\tTest-rmse:31.5439\n",
      "[284]\tTest-rmse:31.5426\n",
      "[285]\tTest-rmse:31.541\n",
      "[286]\tTest-rmse:31.5398\n",
      "[287]\tTest-rmse:31.5384\n",
      "[288]\tTest-rmse:31.5361\n",
      "[289]\tTest-rmse:31.5347\n",
      "[290]\tTest-rmse:31.5324\n",
      "[291]\tTest-rmse:31.5308\n",
      "[292]\tTest-rmse:31.5288\n",
      "[293]\tTest-rmse:31.5271\n",
      "[294]\tTest-rmse:31.5254\n",
      "[295]\tTest-rmse:31.5238\n",
      "[296]\tTest-rmse:31.5223\n",
      "[297]\tTest-rmse:31.5209\n",
      "[298]\tTest-rmse:31.5202\n",
      "[299]\tTest-rmse:31.5184\n",
      "[300]\tTest-rmse:31.5171\n",
      "[301]\tTest-rmse:31.5158\n",
      "[302]\tTest-rmse:31.5145\n",
      "[303]\tTest-rmse:31.5132\n",
      "[304]\tTest-rmse:31.5122\n",
      "[305]\tTest-rmse:31.5109\n",
      "[306]\tTest-rmse:31.51\n",
      "[307]\tTest-rmse:31.5087\n",
      "[308]\tTest-rmse:31.5076\n",
      "[309]\tTest-rmse:31.5064\n",
      "[310]\tTest-rmse:31.5053\n",
      "[311]\tTest-rmse:31.5042\n",
      "[312]\tTest-rmse:31.5032\n",
      "[313]\tTest-rmse:31.5022\n",
      "[314]\tTest-rmse:31.5013\n",
      "[315]\tTest-rmse:31.5003\n",
      "[316]\tTest-rmse:31.4994\n",
      "[317]\tTest-rmse:31.4985\n",
      "[318]\tTest-rmse:31.4975\n",
      "[319]\tTest-rmse:31.4967\n",
      "[320]\tTest-rmse:31.4958\n",
      "[321]\tTest-rmse:31.4955\n",
      "[322]\tTest-rmse:31.4944\n",
      "[323]\tTest-rmse:31.4934\n",
      "[324]\tTest-rmse:31.4929\n",
      "[325]\tTest-rmse:31.492\n",
      "[326]\tTest-rmse:31.4915\n",
      "[327]\tTest-rmse:31.4908\n",
      "[328]\tTest-rmse:31.4896\n",
      "[329]\tTest-rmse:31.4895\n",
      "[330]\tTest-rmse:31.4889\n",
      "[331]\tTest-rmse:31.4882\n",
      "[332]\tTest-rmse:31.4878\n",
      "[333]\tTest-rmse:31.4867\n",
      "[334]\tTest-rmse:31.4857\n",
      "[335]\tTest-rmse:31.4847\n",
      "[336]\tTest-rmse:31.484\n",
      "[337]\tTest-rmse:31.483\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[338]\tTest-rmse:31.4818\n",
      "[339]\tTest-rmse:31.4813\n",
      "[340]\tTest-rmse:31.4806\n",
      "[341]\tTest-rmse:31.4798\n",
      "[342]\tTest-rmse:31.4794\n",
      "[343]\tTest-rmse:31.4784\n",
      "[344]\tTest-rmse:31.478\n",
      "[345]\tTest-rmse:31.477\n",
      "[346]\tTest-rmse:31.4766\n",
      "[347]\tTest-rmse:31.4758\n",
      "[348]\tTest-rmse:31.4754\n",
      "[349]\tTest-rmse:31.475\n",
      "[350]\tTest-rmse:31.4746\n",
      "[351]\tTest-rmse:31.4744\n",
      "[352]\tTest-rmse:31.4735\n",
      "[353]\tTest-rmse:31.4733\n",
      "[354]\tTest-rmse:31.473\n",
      "[355]\tTest-rmse:31.4724\n",
      "[356]\tTest-rmse:31.4719\n",
      "[357]\tTest-rmse:31.4715\n",
      "[358]\tTest-rmse:31.471\n",
      "[359]\tTest-rmse:31.4707\n",
      "[360]\tTest-rmse:31.4695\n",
      "[361]\tTest-rmse:31.4692\n",
      "[362]\tTest-rmse:31.468\n",
      "[363]\tTest-rmse:31.4678\n",
      "[364]\tTest-rmse:31.4673\n",
      "[365]\tTest-rmse:31.4672\n",
      "[366]\tTest-rmse:31.467\n",
      "[367]\tTest-rmse:31.4672\n",
      "[368]\tTest-rmse:31.4672\n",
      "[369]\tTest-rmse:31.467\n",
      "[370]\tTest-rmse:31.4672\n",
      "[371]\tTest-rmse:31.467\n",
      "[372]\tTest-rmse:31.4669\n",
      "[373]\tTest-rmse:31.4672\n",
      "[374]\tTest-rmse:31.467\n",
      "[375]\tTest-rmse:31.4666\n",
      "[376]\tTest-rmse:31.4665\n",
      "[377]\tTest-rmse:31.4668\n",
      "[378]\tTest-rmse:31.4665\n",
      "[379]\tTest-rmse:31.4666\n",
      "[380]\tTest-rmse:31.4671\n",
      "[381]\tTest-rmse:31.4671\n",
      "[382]\tTest-rmse:31.4673\n",
      "[383]\tTest-rmse:31.4669\n",
      "[384]\tTest-rmse:31.4674\n",
      "[385]\tTest-rmse:31.4673\n",
      "[386]\tTest-rmse:31.4668\n",
      "[387]\tTest-rmse:31.4663\n",
      "[388]\tTest-rmse:31.4664\n",
      "[389]\tTest-rmse:31.4659\n",
      "[390]\tTest-rmse:31.466\n",
      "[391]\tTest-rmse:31.4655\n",
      "[392]\tTest-rmse:31.4652\n",
      "[393]\tTest-rmse:31.465\n",
      "[394]\tTest-rmse:31.4651\n",
      "[395]\tTest-rmse:31.4646\n",
      "[396]\tTest-rmse:31.4652\n",
      "[397]\tTest-rmse:31.4654\n",
      "[398]\tTest-rmse:31.465\n",
      "[399]\tTest-rmse:31.4652\n",
      "[400]\tTest-rmse:31.4646\n",
      "[401]\tTest-rmse:31.4644\n",
      "[402]\tTest-rmse:31.4639\n",
      "[403]\tTest-rmse:31.4641\n",
      "[404]\tTest-rmse:31.4635\n",
      "[405]\tTest-rmse:31.4634\n",
      "[406]\tTest-rmse:31.4631\n",
      "[407]\tTest-rmse:31.4627\n",
      "Best RMSE: 31.46 in 408 rounds\n"
     ]
    }
   ],
   "source": [
    "print(\"Our best params: {}\".format(params)) \n",
    "\n",
    "# Let's train our model with best params that we searched\n",
    "best_model = xgb.train(\n",
    "    params,\n",
    "    dtrain,\n",
    "    num_boost_round=num_boost_round,\n",
    "    evals=[(dtest, \"Test\")],\n",
    "    early_stopping_rounds=20\n",
    ")\n",
    "\n",
    "# Saving the model for re-producing the same results\n",
    "print(\"Best RMSE: {:.2f} in {} rounds\".format(best_model.best_score, best_model.best_iteration+1))\n",
    "best_model.save_model(\"xgb_model.model\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[('count_page', 4493),\n",
       " ('entry_session', 4418),\n",
       " ('entry_page', 4250),\n",
       " ('session_durantion', 2733),\n",
       " ('agent_id', 2561),\n",
       " ('traffic_type', 2070),\n",
       " ('row_num', 1281),\n",
       " ('hour_of_day', 756),\n",
       " ('locale_enc', 628),\n",
       " ('combined_sin', 459),\n",
       " ('week_enc', 338),\n",
       " ('combined_cos', 330),\n",
       " ('traffic_enc', 239)]"
      ]
     },
     "execution_count": 34,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Feature importance sorted by the usefullness of individual features\n",
    "sorted(best_model.get_fscore().items(), key=lambda kv: kv[1], reverse=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Converting the dataset to predict the hits and save it in csv file\n",
    "X_submit = test.drop(['hits', 'day_of_week', 'locale', 'path_id_set', 'is_train'], axis=1)\n",
    "dsubmit = xgb.DMatrix(X_submit)\n",
    "\n",
    "submit_pred = best_model.predict(dsubmit)\n",
    "\n",
    "csv_df = pd.DataFrame({'row_num': test['row_num'], 'hits': submit_pred}, columns=['row_num', 'hits'])\n",
    "csv_df.to_csv('submit.csv', index=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 2",
   "language": "python",
   "name": "python2"
  },
  "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
}