TrainEDTalkinggGrid.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\hdadmin\\Anaconda3\\lib\\site-packages\\sklearn\\cross_validation.py:44: 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": [
    "import gc\n",
    "import time\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "from sklearn.cross_validation import train_test_split\n",
    "import xgboost as xgb\n",
    "from xgboost import plot_importance\n",
    "import matplotlib.pyplot as plt"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "path = 'D:\\\\PythonSamples\\\\Talking\\\\'\n",
    "\n",
    "def timeFeatures(df):\n",
    "    # Make some new features with click_time column\n",
    "    df['datetime'] = pd.to_datetime(df['click_time'])\n",
    "    #df['dow']      = df['datetime'].dt.dayofweek\n",
    "    #df[\"doy\"]      = df[\"datetime\"].dt.dayofyear\n",
    "    df[\"date_day\"]      = df[\"datetime\"].dt.day\n",
    "    df[\"date_hour\"]      = df[\"datetime\"].dt.hour\n",
    "    #df[\"dteom\"]    = df[\"datetime\"].dt.daysinmonth - df[\"datetime\"].dt.day\n",
    "    df.drop(['click_time', 'datetime'], axis=1, inplace=True)\n",
    "    return df"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "start_time = time.time()\n",
    "\n",
    "train_columns = ['ip', 'app', 'device', 'os', 'channel', 'click_time', 'is_attributed']\n",
    "test_columns  = ['ip', 'app', 'device', 'os', 'channel', 'click_time', 'click_id']\n",
    "dtypes = {\n",
    "        'ip'            : 'uint32',\n",
    "        'app'           : 'uint16',\n",
    "        'device'        : 'uint16',\n",
    "        'os'            : 'uint16',\n",
    "        'channel'       : 'uint16',\n",
    "        'is_attributed' : 'uint8',\n",
    "        'click_id'      : 'uint32'\n",
    "        }"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "train_ref = pd.read_csv(path+\"train.csv\", usecols=train_columns, dtype=dtypes)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "test_ref = pd.read_csv(path+\"test.csv\", usecols=test_columns, dtype=dtypes)"
   ]
  },
  {
   "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>ip</th>\n",
       "      <th>app</th>\n",
       "      <th>device</th>\n",
       "      <th>os</th>\n",
       "      <th>channel</th>\n",
       "      <th>is_attributed</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>count</th>\n",
       "      <td>1.849039e+08</td>\n",
       "      <td>1.849039e+08</td>\n",
       "      <td>1.849039e+08</td>\n",
       "      <td>1.849039e+08</td>\n",
       "      <td>1.849039e+08</td>\n",
       "      <td>1.849039e+08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mean</th>\n",
       "      <td>9.087604e+04</td>\n",
       "      <td>1.201131e+01</td>\n",
       "      <td>2.172325e+01</td>\n",
       "      <td>2.267702e+01</td>\n",
       "      <td>2.685789e+02</td>\n",
       "      <td>2.470721e-03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>std</th>\n",
       "      <td>6.952789e+04</td>\n",
       "      <td>1.480521e+01</td>\n",
       "      <td>2.593326e+02</td>\n",
       "      <td>5.525282e+01</td>\n",
       "      <td>1.295882e+02</td>\n",
       "      <td>4.964491e-02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>min</th>\n",
       "      <td>1.000000e+00</td>\n",
       "      <td>0.000000e+00</td>\n",
       "      <td>0.000000e+00</td>\n",
       "      <td>0.000000e+00</td>\n",
       "      <td>0.000000e+00</td>\n",
       "      <td>0.000000e+00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25%</th>\n",
       "      <td>4.024500e+04</td>\n",
       "      <td>3.000000e+00</td>\n",
       "      <td>1.000000e+00</td>\n",
       "      <td>1.300000e+01</td>\n",
       "      <td>1.400000e+02</td>\n",
       "      <td>0.000000e+00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>50%</th>\n",
       "      <td>7.962200e+04</td>\n",
       "      <td>1.200000e+01</td>\n",
       "      <td>1.000000e+00</td>\n",
       "      <td>1.800000e+01</td>\n",
       "      <td>2.580000e+02</td>\n",
       "      <td>0.000000e+00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>75%</th>\n",
       "      <td>1.182470e+05</td>\n",
       "      <td>1.500000e+01</td>\n",
       "      <td>1.000000e+00</td>\n",
       "      <td>1.900000e+01</td>\n",
       "      <td>3.790000e+02</td>\n",
       "      <td>0.000000e+00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>max</th>\n",
       "      <td>3.647780e+05</td>\n",
       "      <td>7.680000e+02</td>\n",
       "      <td>4.227000e+03</td>\n",
       "      <td>9.560000e+02</td>\n",
       "      <td>5.000000e+02</td>\n",
       "      <td>1.000000e+00</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                 ip           app        device            os       channel  \\\n",
       "count  1.849039e+08  1.849039e+08  1.849039e+08  1.849039e+08  1.849039e+08   \n",
       "mean   9.087604e+04  1.201131e+01  2.172325e+01  2.267702e+01  2.685789e+02   \n",
       "std    6.952789e+04  1.480521e+01  2.593326e+02  5.525282e+01  1.295882e+02   \n",
       "min    1.000000e+00  0.000000e+00  0.000000e+00  0.000000e+00  0.000000e+00   \n",
       "25%    4.024500e+04  3.000000e+00  1.000000e+00  1.300000e+01  1.400000e+02   \n",
       "50%    7.962200e+04  1.200000e+01  1.000000e+00  1.800000e+01  2.580000e+02   \n",
       "75%    1.182470e+05  1.500000e+01  1.000000e+00  1.900000e+01  3.790000e+02   \n",
       "max    3.647780e+05  7.680000e+02  4.227000e+03  9.560000e+02  5.000000e+02   \n",
       "\n",
       "       is_attributed  \n",
       "count   1.849039e+08  \n",
       "mean    2.470721e-03  \n",
       "std     4.964491e-02  \n",
       "min     0.000000e+00  \n",
       "25%     0.000000e+00  \n",
       "50%     0.000000e+00  \n",
       "75%     0.000000e+00  \n",
       "max     1.000000e+00  "
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_ref.describe()"
   ]
  },
  {
   "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>click_id</th>\n",
       "      <th>ip</th>\n",
       "      <th>app</th>\n",
       "      <th>device</th>\n",
       "      <th>os</th>\n",
       "      <th>channel</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>count</th>\n",
       "      <td>1.879047e+07</td>\n",
       "      <td>1.879047e+07</td>\n",
       "      <td>1.879047e+07</td>\n",
       "      <td>1.879047e+07</td>\n",
       "      <td>1.879047e+07</td>\n",
       "      <td>1.879047e+07</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mean</th>\n",
       "      <td>9.395234e+06</td>\n",
       "      <td>6.306921e+04</td>\n",
       "      <td>1.221480e+01</td>\n",
       "      <td>1.730513e+00</td>\n",
       "      <td>1.873312e+01</td>\n",
       "      <td>2.648059e+02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>std</th>\n",
       "      <td>5.424341e+06</td>\n",
       "      <td>3.688597e+04</td>\n",
       "      <td>1.164924e+01</td>\n",
       "      <td>2.597038e+01</td>\n",
       "      <td>1.135059e+01</td>\n",
       "      <td>1.355254e+02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>min</th>\n",
       "      <td>0.000000e+00</td>\n",
       "      <td>0.000000e+00</td>\n",
       "      <td>0.000000e+00</td>\n",
       "      <td>0.000000e+00</td>\n",
       "      <td>0.000000e+00</td>\n",
       "      <td>0.000000e+00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25%</th>\n",
       "      <td>4.697617e+06</td>\n",
       "      <td>3.155800e+04</td>\n",
       "      <td>3.000000e+00</td>\n",
       "      <td>1.000000e+00</td>\n",
       "      <td>1.300000e+01</td>\n",
       "      <td>1.350000e+02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>50%</th>\n",
       "      <td>9.395234e+06</td>\n",
       "      <td>6.393600e+04</td>\n",
       "      <td>1.200000e+01</td>\n",
       "      <td>1.000000e+00</td>\n",
       "      <td>1.800000e+01</td>\n",
       "      <td>2.360000e+02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>75%</th>\n",
       "      <td>1.409285e+07</td>\n",
       "      <td>9.531600e+04</td>\n",
       "      <td>1.800000e+01</td>\n",
       "      <td>1.000000e+00</td>\n",
       "      <td>1.900000e+01</td>\n",
       "      <td>4.010000e+02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>max</th>\n",
       "      <td>1.879047e+07</td>\n",
       "      <td>1.264130e+05</td>\n",
       "      <td>5.210000e+02</td>\n",
       "      <td>3.031000e+03</td>\n",
       "      <td>6.040000e+02</td>\n",
       "      <td>4.980000e+02</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "           click_id            ip           app        device            os  \\\n",
       "count  1.879047e+07  1.879047e+07  1.879047e+07  1.879047e+07  1.879047e+07   \n",
       "mean   9.395234e+06  6.306921e+04  1.221480e+01  1.730513e+00  1.873312e+01   \n",
       "std    5.424341e+06  3.688597e+04  1.164924e+01  2.597038e+01  1.135059e+01   \n",
       "min    0.000000e+00  0.000000e+00  0.000000e+00  0.000000e+00  0.000000e+00   \n",
       "25%    4.697617e+06  3.155800e+04  3.000000e+00  1.000000e+00  1.300000e+01   \n",
       "50%    9.395234e+06  6.393600e+04  1.200000e+01  1.000000e+00  1.800000e+01   \n",
       "75%    1.409285e+07  9.531600e+04  1.800000e+01  1.000000e+00  1.900000e+01   \n",
       "max    1.879047e+07  1.264130e+05  5.210000e+02  3.031000e+03  6.040000e+02   \n",
       "\n",
       "            channel  \n",
       "count  1.879047e+07  \n",
       "mean   2.648059e+02  \n",
       "std    1.355254e+02  \n",
       "min    0.000000e+00  \n",
       "25%    1.350000e+02  \n",
       "50%    2.360000e+02  \n",
       "75%    4.010000e+02  \n",
       "max    4.980000e+02  "
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "test_ref.describe()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "50"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Drop the IP and the columns from target\n",
    "train_ref.drop(['is_attributed'], axis=1, inplace=True)\n",
    "gc.collect()\n",
    "#nrow_train = train.shape[0]\n",
    "merge_ref = pd.concat([train_ref, test_ref])\n",
    "del train_ref, test_ref\n",
    "gc.collect()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "collapsed": true,
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "#train_ref = timeFeatures(train_ref)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "ip_appcount = merge_ref.groupby(['ip']).app.nunique().reset_index()\n",
    "ip_appcount.columns = ['ip', 'ip_appcounts']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "ip_oscount = merge_ref.groupby(['ip']).os.nunique().reset_index()\n",
    "ip_oscount.columns = ['ip', 'ip_oscounts']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "app_ipcount = merge_ref.groupby(['app'])['ip'].count().reset_index()\n",
    "app_ipcount.columns = ['app', 'app_clickip']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "app_app_oscount = merge_ref.groupby(['app','os'])['channel'].count().reset_index()\n",
    "app_app_oscount.columns = ['app', 'os', 'app_os_clicks']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "app_app_devcount = merge_ref.groupby(['app','device'])['channel'].count().reset_index()\n",
    "app_app_devcount.columns = ['app', 'device', 'app_device_click']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Count the number of clicks by ip\n",
    "ip_count = merge_ref.groupby(['ip'])['channel'].count().reset_index()\n",
    "ip_count.columns = ['ip', 'clicks_by_ip']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "#app_count = merge_ref.groupby(['app'])['channel'].count().reset_index()\n",
    "#app_count.columns = ['app', 'clicks_by_app']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "app_os_dev_count = merge_ref.groupby(['app','os','device'])['channel'].count().reset_index()\n",
    "app_os_dev_count.columns = ['app','os','device','app_os_dev_count']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "28"
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "del merge_ref\n",
    "gc.collect()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "train_rank = pd.read_csv(path+\"train.csv\") #, usecols=train_columns, dtype=dtypes)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "#train_rank.isnull().sum()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "#train_rank.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "apprank = train_rank.groupby(['app'])['is_attributed'].sum().reset_index().astype('int')\n",
    "apprank.columns = ['app', 'app_sum']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "rnk = apprank.loc[apprank['app_sum'] != 0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\hdadmin\\Anaconda3\\lib\\site-packages\\ipykernel_launcher.py:1: SettingWithCopyWarning: \n",
      "A value is trying to be set on a copy of a slice from a DataFrame.\n",
      "Try using .loc[row_indexer,col_indexer] = value instead\n",
      "\n",
      "See the caveats in the documentation: http://pandas.pydata.org/pandas-docs/stable/indexing.html#indexing-view-versus-copy\n",
      "  \"\"\"Entry point for launching an IPython kernel.\n"
     ]
    }
   ],
   "source": [
    "rnk['app_rank'] = rnk['app_sum'].rank(ascending=False).astype('int')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\hdadmin\\Anaconda3\\lib\\site-packages\\ipykernel_launcher.py:1: SettingWithCopyWarning: \n",
      "A value is trying to be set on a copy of a slice from a DataFrame\n",
      "\n",
      "See the caveats in the documentation: http://pandas.pydata.org/pandas-docs/stable/indexing.html#indexing-view-versus-copy\n",
      "  \"\"\"Entry point for launching an IPython kernel.\n"
     ]
    }
   ],
   "source": [
    "rnk.drop(['app_sum'], axis=1, inplace=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "45"
      ]
     },
     "execution_count": 26,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "del apprank\n",
    "del train_rank\n",
    "gc.collect()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "train_check_1 = pd.read_csv(path+\"train.csv\", usecols=train_columns, dtype=dtypes)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "test = pd.read_csv(path+\"test.csv\", usecols=test_columns, dtype=dtypes)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "apps = test['app'].unique()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "train_check_1 = train_check_1.loc[train_check_1['app'].isin(apps)]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "devices = test['device'].unique()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "train_check_1 = train_check_1.loc[train_check_1['device'].isin(devices)]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "oss = test['os'].unique()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "train_check_1 = train_check_1.loc[train_check_1['os'].isin(oss)]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "channels = test['channel'].unique()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "train_check_1 = train_check_1.loc[train_check_1['channel'].isin(channels)]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "train_1 = train_check_1.loc[train_check_1[\"is_attributed\"] == 1]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "97"
      ]
     },
     "execution_count": 38,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "del apps,devices,oss,channels\n",
    "gc.collect()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "#train_check = pd.read_csv(path+\"train.csv\", skiprows=range(1,127903891), nrows=57000000, usecols=train_columns, dtype=dtypes)\n",
    "#import random\n",
    "#train = pd.read_csv(path+\"distinct_train.csv\", usecols=train_columns, dtype=dtypes, skiprows=lambda i: i>0 and random.random() > fraction_val)                     "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "train_0 = train_check_1.loc[train_check_1[\"is_attributed\"] == 0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "24"
      ]
     },
     "execution_count": 41,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "del train_check_1\n",
    "gc.collect()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "part1 = train_0.iloc[0:18490381]\n",
    "part2 = train_0.iloc[18490381:36980761]\n",
    "part3 = train_0.iloc[36980761:55471141]\n",
    "part4 = train_0.iloc[55471141:73961521]\n",
    "part5 = train_0.iloc[73961521:92451902]\n",
    "part6 = train_0.iloc[92451901:110942282]\n",
    "part7 = train_0.iloc[110942282:129432662]\n",
    "part8 = train_0.iloc[129432662:147923042]\n",
    "part9 = train_0.iloc[147923042:166413422]\n",
    "part10 = train_0.iloc[166413422:184903890]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "13"
      ]
     },
     "execution_count": 43,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "del train_0\n",
    "gc.collect()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "part1 = part1.sample(n=100000)\n",
    "part2 = part2.sample(n=100000)\n",
    "part3 = part3.sample(n=100000)\n",
    "part4 = part4.sample(n=100000)\n",
    "part5 = part5.sample(n=100000)\n",
    "part6 = part6.sample(n=100000)\n",
    "part7 = part7.sample(n=100000)\n",
    "part8 = part8.sample(n=100000)\n",
    "part9 = part9.sample(n=100000)\n",
    "part10 = part10.sample(n=100000)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0"
      ]
     },
     "execution_count": 45,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "gc.collect()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0"
      ]
     },
     "execution_count": 46,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train = pd.concat([train_1, part1, part2, part3, part4, part5, part6, part7, part8, part9, part10])\n",
    "del  train_1, part1, part2, part3, part4, part5, part6, part7, part8, part9, part10\n",
    "gc.collect()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "#train_0 = train_0.tail(57000000)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "#del train_check_1\n",
    "#gc.collect()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "#fraction_val = 3000000/57000000\n",
    "#print(fraction_val)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "#train_0f = train_0.sample(frac=fraction_val)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "#train = pd.concat([train_1, train_0f])\n",
    "#del  train_0f, train_1\n",
    "#gc.collect()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "train = train.sample(frac=1).reset_index(drop=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "42"
      ]
     },
     "execution_count": 53,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Drop the IP and the columns from target\n",
    "y = train['is_attributed']\n",
    "train.drop(['is_attributed'], axis=1, inplace=True)\n",
    "\n",
    "# Drop IP and ID from test rows\n",
    "sub = pd.DataFrame()\n",
    "sub['click_id'] = test['click_id'].astype('int')\n",
    "#test.drop(['click_id'], axis=1, inplace=True)\n",
    "#gc.collect()\n",
    "\n",
    "#nrow_train = train.shape[0]\n",
    "#merge = pd.concat([train, test])\n",
    "del test\n",
    "gc.collect()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "train = timeFeatures(train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "train = pd.merge(train, ip_appcount, on='ip', how='left', sort=False)\n",
    "train['ip_appcounts'].fillna(0, inplace=True)\n",
    "train['ip_appcounts'] = train['ip_appcounts'].astype('uint16')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "train = pd.merge(train, ip_oscount, on='ip', how='left', sort=False)\n",
    "train['ip_oscounts'].fillna(0, inplace=True)\n",
    "train['ip_oscounts'] = train['ip_oscounts'].astype('uint16')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "train = pd.merge(train, app_ipcount, on='app', how='left', sort=False)\n",
    "train['app_clickip'].fillna(0, inplace=True)\n",
    "train['app_clickip'] = train['app_clickip'].astype('uint16')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "train = pd.merge(train, app_app_oscount, on=['app','os'], how='left', sort=False)\n",
    "train['app_os_clicks'].fillna(0, inplace=True)\n",
    "train['app_os_clicks'] = train['app_os_clicks'].astype('uint16')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 59,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "train = pd.merge(train, app_app_devcount, on=['app','device'], how='left', sort=False)\n",
    "train['app_device_click'].fillna(0, inplace=True)\n",
    "train['app_device_click'] = train['app_device_click'].astype('uint16')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "train = pd.merge(train, ip_count, on='ip', how='left', sort=False)\n",
    "train['clicks_by_ip'].fillna(0, inplace=True)\n",
    "train['clicks_by_ip'] = train['clicks_by_ip'].astype('uint16')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 61,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "#train = pd.merge(train, app_count, on=['app'], how='left', sort=False)\n",
    "#train['clicks_by_app'] = train['clicks_by_app'].astype('uint16')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 62,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "train = pd.merge(train, app_os_dev_count, on=['app','os','device'], how='left', sort=False)\n",
    "train['app_os_dev_count'].fillna(0, inplace=True)\n",
    "train['app_os_dev_count'] =train['app_os_dev_count'].astype('uint16')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 63,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "train = pd.merge(train, rnk, on=['app'], how='left', sort=False)\n",
    "train['app_rank'].fillna(0, inplace=True)\n",
    "train['app_rank'] = train['app_rank'].astype('uint16')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 64,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "84"
      ]
     },
     "execution_count": 64,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train.drop('ip', axis=1, inplace=True)\n",
    "train.drop('date_day', axis=1, inplace=True)\n",
    "#train.drop('os', axis=1, inplace=True)\n",
    "gc.collect()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 65,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "#train.drop('os', axis=1, inplace=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 66,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "import seaborn as sb\n",
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 67,
   "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>app</th>\n",
       "      <th>device</th>\n",
       "      <th>os</th>\n",
       "      <th>channel</th>\n",
       "      <th>date_hour</th>\n",
       "      <th>ip_appcounts</th>\n",
       "      <th>ip_oscounts</th>\n",
       "      <th>app_clickip</th>\n",
       "      <th>app_os_clicks</th>\n",
       "      <th>app_device_click</th>\n",
       "      <th>clicks_by_ip</th>\n",
       "      <th>app_os_dev_count</th>\n",
       "      <th>app_rank</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>12</td>\n",
       "      <td>1</td>\n",
       "      <td>19</td>\n",
       "      <td>326</td>\n",
       "      <td>20</td>\n",
       "      <td>132</td>\n",
       "      <td>114</td>\n",
       "      <td>8542</td>\n",
       "      <td>12172</td>\n",
       "      <td>47068</td>\n",
       "      <td>16902</td>\n",
       "      <td>53250</td>\n",
       "      <td>19</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>23</td>\n",
       "      <td>1</td>\n",
       "      <td>19</td>\n",
       "      <td>153</td>\n",
       "      <td>22</td>\n",
       "      <td>109</td>\n",
       "      <td>110</td>\n",
       "      <td>22839</td>\n",
       "      <td>44020</td>\n",
       "      <td>1534</td>\n",
       "      <td>47421</td>\n",
       "      <td>40992</td>\n",
       "      <td>112</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>20</td>\n",
       "      <td>1</td>\n",
       "      <td>13</td>\n",
       "      <td>259</td>\n",
       "      <td>23</td>\n",
       "      <td>193</td>\n",
       "      <td>138</td>\n",
       "      <td>33445</td>\n",
       "      <td>18452</td>\n",
       "      <td>54810</td>\n",
       "      <td>40368</td>\n",
       "      <td>29471</td>\n",
       "      <td>16</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>10</td>\n",
       "      <td>1</td>\n",
       "      <td>19</td>\n",
       "      <td>377</td>\n",
       "      <td>2</td>\n",
       "      <td>52</td>\n",
       "      <td>44</td>\n",
       "      <td>24588</td>\n",
       "      <td>52695</td>\n",
       "      <td>20011</td>\n",
       "      <td>5982</td>\n",
       "      <td>52154</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>3</td>\n",
       "      <td>1</td>\n",
       "      <td>11</td>\n",
       "      <td>409</td>\n",
       "      <td>3</td>\n",
       "      <td>50</td>\n",
       "      <td>43</td>\n",
       "      <td>2444</td>\n",
       "      <td>60257</td>\n",
       "      <td>63603</td>\n",
       "      <td>3813</td>\n",
       "      <td>34563</td>\n",
       "      <td>9</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   app  device  os  channel  date_hour  ip_appcounts  ip_oscounts  \\\n",
       "0   12       1  19      326         20           132          114   \n",
       "1   23       1  19      153         22           109          110   \n",
       "2   20       1  13      259         23           193          138   \n",
       "3   10       1  19      377          2            52           44   \n",
       "4    3       1  11      409          3            50           43   \n",
       "\n",
       "   app_clickip  app_os_clicks  app_device_click  clicks_by_ip  \\\n",
       "0         8542          12172             47068         16902   \n",
       "1        22839          44020              1534         47421   \n",
       "2        33445          18452             54810         40368   \n",
       "3        24588          52695             20011          5982   \n",
       "4         2444          60257             63603          3813   \n",
       "\n",
       "   app_os_dev_count  app_rank  \n",
       "0             53250        19  \n",
       "1             40992       112  \n",
       "2             29471        16  \n",
       "3             52154         4  \n",
       "4             34563         9  "
      ]
     },
     "execution_count": 67,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 68,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "test = pd.read_csv(path+\"test.csv\", usecols=test_columns, dtype=dtypes)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 69,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "test = timeFeatures(test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 70,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "test = pd.merge(test, ip_appcount, on='ip', how='left', sort=False)\n",
    "test = pd.merge(test, ip_oscount, on='ip', how='left', sort=False)\n",
    "test = pd.merge(test, app_ipcount, on='app', how='left', sort=False)\n",
    "test = pd.merge(test, app_app_oscount, on=['app','os'], how='left', sort=False)\n",
    "test = pd.merge(test, app_app_devcount, on=['app','device'], how='left', sort=False)\n",
    "test = pd.merge(test, ip_count, on='ip', how='left', sort=False)\n",
    "#test = pd.merge(test, app_count, on=['app'], how='left', sort=False)\n",
    "test = pd.merge(test, app_os_dev_count, on=['app','os','device'], how='left', sort=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 71,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "test['ip_appcounts'].fillna(0, inplace=True)\n",
    "test['ip_oscounts'].fillna(0, inplace=True)\n",
    "test['app_clickip'].fillna(0, inplace=True)\n",
    "test['app_os_clicks'].fillna(0, inplace=True)\n",
    "test['app_device_click'].fillna(0, inplace=True)\n",
    "test['clicks_by_ip'].fillna(0, inplace=True)\n",
    "#test['clicks_by_app'].fillna(0, inplace=True)\n",
    "test['app_os_dev_count'].fillna(0, inplace=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 72,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "test['ip_appcounts'] = test['ip_appcounts'].astype('uint16')\n",
    "test['ip_oscounts'] = test['ip_oscounts'].astype('uint16')\n",
    "test['app_clickip'] = test['app_clickip'].astype('uint16')\n",
    "test['app_os_clicks'] = test['app_os_clicks'].astype('uint16')\n",
    "test['app_device_click'] = test['app_device_click'].astype('uint16')\n",
    "test['clicks_by_ip'] = test['clicks_by_ip'].astype('uint16')\n",
    "#test['clicks_by_app'] = test['clicks_by_app'].astype('uint16')\n",
    "test['app_os_dev_count'] = test['app_os_dev_count'].astype('uint16')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 73,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "test = pd.merge(test, rnk, on=['app'], how='left', sort=False)\n",
    "test['app_rank'].fillna(0, inplace=True)\n",
    "test['app_rank'] = test['app_rank'].astype('uint16')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 74,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "196"
      ]
     },
     "execution_count": 74,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "del ip_appcount,app_ipcount,ip_count,app_app_oscount,app_app_devcount\n",
    "del rnk\n",
    "gc.collect()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 75,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "test.drop(['click_id', 'ip'], axis=1, inplace=True)\n",
    "test.drop('date_day', axis=1, inplace=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 76,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "#test.drop('os', axis=1, inplace=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 77,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "#train.drop(['app_clickip','app_os_clicks','app_device_click','clicks_by_ip'], axis=1, inplace=True)\n",
    "#test.drop(['app_clickip','app_os_clicks','app_device_click','clicks_by_ip'], axis=1, inplace=True)\n",
    "\n",
    "#train.drop(['channel'], axis=1, inplace=True)\n",
    "#test.drop(['channel'], axis=1, inplace=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 78,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "#train.drop(['date_hour', 'app_os_dev_count'], axis=1, inplace=True)\n",
    "#test.drop(['date_hour', 'app_os_dev_count'], axis=1, inplace=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 79,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "#train.drop(['device','app_clickip'], axis=1, inplace=True)\n",
    "#test.drop(['device','app_clickip'], axis=1, inplace=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 80,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "train.drop(['device','ip_appcounts'], axis=1, inplace=True)\n",
    "test.drop(['device','ip_appcounts'], axis=1, inplace=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 81,
   "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>app</th>\n",
       "      <th>os</th>\n",
       "      <th>channel</th>\n",
       "      <th>date_hour</th>\n",
       "      <th>ip_oscounts</th>\n",
       "      <th>app_clickip</th>\n",
       "      <th>app_os_clicks</th>\n",
       "      <th>app_device_click</th>\n",
       "      <th>clicks_by_ip</th>\n",
       "      <th>app_os_dev_count</th>\n",
       "      <th>app_rank</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>9</td>\n",
       "      <td>3</td>\n",
       "      <td>107</td>\n",
       "      <td>4</td>\n",
       "      <td>44</td>\n",
       "      <td>62860</td>\n",
       "      <td>28887</td>\n",
       "      <td>44385</td>\n",
       "      <td>1134</td>\n",
       "      <td>14506</td>\n",
       "      <td>6</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>9</td>\n",
       "      <td>3</td>\n",
       "      <td>466</td>\n",
       "      <td>4</td>\n",
       "      <td>67</td>\n",
       "      <td>62860</td>\n",
       "      <td>28887</td>\n",
       "      <td>44385</td>\n",
       "      <td>23273</td>\n",
       "      <td>14506</td>\n",
       "      <td>6</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>21</td>\n",
       "      <td>19</td>\n",
       "      <td>128</td>\n",
       "      <td>4</td>\n",
       "      <td>60</td>\n",
       "      <td>60834</td>\n",
       "      <td>53180</td>\n",
       "      <td>63605</td>\n",
       "      <td>15544</td>\n",
       "      <td>47470</td>\n",
       "      <td>58</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>15</td>\n",
       "      <td>13</td>\n",
       "      <td>111</td>\n",
       "      <td>4</td>\n",
       "      <td>57</td>\n",
       "      <td>64259</td>\n",
       "      <td>64622</td>\n",
       "      <td>59348</td>\n",
       "      <td>12827</td>\n",
       "      <td>16402</td>\n",
       "      <td>17</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>12</td>\n",
       "      <td>13</td>\n",
       "      <td>328</td>\n",
       "      <td>4</td>\n",
       "      <td>33</td>\n",
       "      <td>8542</td>\n",
       "      <td>36788</td>\n",
       "      <td>47068</td>\n",
       "      <td>1441</td>\n",
       "      <td>36875</td>\n",
       "      <td>19</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   app  os  channel  date_hour  ip_oscounts  app_clickip  app_os_clicks  \\\n",
       "0    9   3      107          4           44        62860          28887   \n",
       "1    9   3      466          4           67        62860          28887   \n",
       "2   21  19      128          4           60        60834          53180   \n",
       "3   15  13      111          4           57        64259          64622   \n",
       "4   12  13      328          4           33         8542          36788   \n",
       "\n",
       "   app_device_click  clicks_by_ip  app_os_dev_count  app_rank  \n",
       "0             44385          1134             14506         6  \n",
       "1             44385         23273             14506         6  \n",
       "2             63605         15544             47470        58  \n",
       "3             59348         12827             16402        17  \n",
       "4             47068          1441             36875        19  "
      ]
     },
     "execution_count": 81,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "test.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 82,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x1af0e709b0>"
      ]
     },
     "execution_count": 82,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAawAAAFDCAYAAABm2jr7AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAAIABJREFUeJzt3XeYZFW1/vFvdzMDMzAEQRzAgKR3QFGQHBRQQcWEijIC\nKiCXYCIZCHoRL6DXn4KgCAxBQEWSF0VFUEHiDGICQZglSFIQkDykSd2/P/Yupmg71HTv012n+/08\nTz1dfapqndM9PbVq77PPWh09PT2YmZm1u87RPgAzM7NWOGGZmVktOGGZmVktOGGZmVktOGGZmVkt\nLDHaB2CL7NexeqVLNk94+pYqwwNw/3PVxl9lmQmVxr/vqXmVxgd42eRq/9vd9ki1/whz5i6sND7A\n6162dKXxr73vyUrj7/TMrErjA0zc4gMdw42xOO85p/TcM+z9DZdHWGZmVgseYZmZjVMTO0d90LRY\nnLDMzMaprnrlKycsM7PxqqujXhnLCcvMbJzyCMvMzGrBIywzM6sFj7DMzKwWJniEZWZmdeApwTFO\n0rLA6cDywKrAScAuwGxgGtCRv58GHAF0A1OBGRFx0mgcs5lZX+o2JehKF4tvLeC8iNgB2AE4OG+f\nGRHbAucDh+dtqwHvATYHDpK08ggfq5lZv7o6Olq+tQOPsBbfQ8CBkt4PPAU0ittdmb/OBN7buB8R\ncwEk3QqsCTw8gsdqZtYvj7DGvkOAWRGxO3AhaQoQYKP8dSvgr/n+BpK6JE0GXgPcMaJHamY2gImd\nHS3f2oFHWIvvZ8C3JU0HngAWAEsCe0g6GHgG+AiwPmn09UtgReDoiHhkdA7ZzOw/tctUX6ucsBZT\nRPwWeG3zNklXAYdFxOymbQC3R8T0ET1AM7MW1W1K0AnLzGyccsIah/LqwN7brgKuGuljMTNrlacE\nzcysFjzCMjOzWii1+k9SJ/Bd4PXAXGDviLiz6fHdSCusFwJnRsTJQ9mPl7WbmY1TBS8c3glYKiK2\nAA4Fvtnr8W8AbyVd9nOIpBWGcrweYbWRE56+pdL4ByyzfqXxAb7xzO2Vxu/o6ak0/kqTqv8v0VXx\nNS3TVpxUafxq/wWSJSr+HW222rKVxp87acdK4wNMLBCj4JTg1sBlABFxg6SNez3+F2A50mVAHQzx\nz8gjLDOzcargCGtZ4Mmm7xdKav70dyvwR1JRhZ9HxBNDOV4nLDOzcaqzo6Pl2yCeAqY0h46IBQCS\nXge8E3g1sDqwsqQPDul4h/IiMzOrv46ujpZvg7ge2BFA0uZA8/mNJ4HngOciYiGpnqrPYZmZWeu6\nJnaVCnUxsL2kmaRzVHtK2hVYJiJmSDoVuE7SPODvwFlD2YkTlpnZONXCyKklEdEN7Ndr8+ymx08B\nThnufpywzMzGqc6aXTnshGVmNk51dNZrGYMTViGSJgDfA9YAuoDjSG1FPgZ0A7+PiM+M3hGamb1Y\n3UZY9Uqv7W1f4N8RsSXpiu6jgc8Bn8pXf9/e67oEM7NR1Tmxq+VbO3DCKmdd4BqAiJgD3AbsAXxS\n0tXAq1jUndjMbNR1dnW0fGsHTljl3A68EUDSFFLH4enAfhGxDbAhsOXoHZ6Z2Yt1dHa0fGsHnqIq\nZwZwmqTrgEnAUcAE4FpJc4D7gd+N4vGZmb1IZ1e9xixOWIVExDzSAoveTh/pYzEza0Wp67BGihOW\nmdk45YRlZma1ULA004hwwjIzG6c622QxRaucsMzMxqkOL7owM7M6aJfrq1rlhNVG7n+u2vhVt68H\n+OzS61Ya/4Snbqo0/rLPDakR6mLp6az2v92k+dX+IXVPmFRpfICeiZMrjf/yOfdVGn/+pLUrjV+K\nF12YmVkteErQzMxqoWuCE5aZmdWAK12YmVkt+ByWmZnVgs9hmZlZLbjjcE1I2gOYFhGHjtD+vgw8\nGBGnjMT+zMwG0zWxXimgXkdrZmbFeEqwTUmaBHyP1Pl3InARsLmkXwEvBU6OiBmSdgY+Sepl1QO8\nD3gt8AVgHrAGcF5EHCPpLGAusDqwCrBHRPxJ0geBg4GFwHUjNYozM1scdZsSrNfRDs9+wD0RsQWp\nE/BzwHzgbaSkdGB+3jrAOyNia1Kb+7fl7a8CPgBsDny+Ke69EfE24NvAPpJeQmre+JYcYzVJ21f6\nk5mZDUFHV1fLt3YwnhKWgFkAEXEH8ATwp4joAR4EGrVgHgbOlvQ94HWkkRbALRGxICKeISW7hj/n\nr/8AlgLWIo3YLpV0FbAesGZVP5SZ2VB1dHW2fGsH7XEUI+N2YBMASWsAx5Km/F4gaTnS6Gg6sDcp\nMTUuVHjRc5v03n43KXltHxHbkkZeNwz/8M3Myurs7Gz51g7GzTks4FTgTElXA13AccBKvZ7zFHA9\naSS2AHgcWJWUhFoSEf+WdBxwtaQu4B7ggmEfvZlZYZ01WyXY0dPT38DBRtpdj8yp9B9j5cnV/3HW\nvVp75xio1t7pau2DWuKxiqu1r1x9tfalJi897DIV9x+1b8vvOasdeeqol8WoV3o1M7Ni2uXcVKuc\nsMzMxiknLDMzqwVXazczs1ronFCvFFCvozUzs2I8JWhDtsoyEwZ/0jB0jMCK0KpX8R2w7AaVxj/x\n39dVGh+gZ8llKo2/sOIVdvOXqH6V4MRnH600/oKXVnstf8fC+ZXGL6VupZmcsMzMxqnONim51Con\nLDOzccpTgmZmVgulEpakTuC7wOtJHSz2jog7+3jeDOCxoXawqFd6NTOzYjonLNHybRA7AUvlbhiH\nAt/s/QRJ+wLrD+t4h/NiMzOrr4LV2rcGLgOIiBuAjZsflLQlsBmppuuQOWGZmY1THZ2dLd8GsSzw\nZNP3CyUtASBpFeBI4FPDPd4xk7AkLSXpngEe30fSYq0bl7S6JLcGMbMxqaOzq+XbIJ4CpjR93xkR\nC/L9D5I6Y1xKmi7cVdIeQzne8bTo4nDgHFKXYTMzGzwRtep64N3ABZI2B25pPBARJwInAuRENS0i\nzhrKTmqdsCQtA/wQWAG4M2/bhjT87ASWAXYF3ghMBc4DdpL01bytCzguIi4cYDcvlfQTYBXgLxHx\nX5JWB84k/f56gM9ExM2SHoyIqfk4zgNOAVYH9srHc2REXFHuN2BmNgzlLhy+GNhe0kxS09s9Je0K\nLBMRM0rtpNYJC9gPuDUijpC0GfBm4DXA7hHxgKTDgQ9GxDGSvgRMl/QO4NURsbWkpYAbJP06Ivpr\nhLQssCdpfvZOSSsD3wBOiIifStoAOINeJxl7eTwi3lvkJzYzK6RjwsQicSKim/R+3Gx2H887azj7\nqfs5rHWAGwEi4nek6b77gRMlnQVsB/Q+b7U+sJGkq0irWiaQRkH9uSsiHs//IA8Dk4F1gWvyfm8C\nXtHH65qbncXi/FBmZiOis6v1Wxuoe8K6DdgCQNKGpORzGrBnROwBPMCixNFN+nlnA7+NiG1JI7IL\ngL8PsI++CvDdTppSJI+wHszbJ0haRtJE0kivoXtxfzAzs6oVXCU4Iuo+JXgKcI6k60iJaC7wf8C1\nkp4BHgJWzc+9lrRKZTtgW0nXks5xXRwRcxZzv58FTpP0WVKS/Hje/i3gBuAu4N4h/1RmZiOhTUZO\nreroGYEK3taa555/vtJ/jJGo1t6xYG6l8V2tvZUdVDugHwvV2nsmLVdpfLoXVhsfWHLK8h2DP2tg\nz//qjJbfFJba4ePD3t9w1X2EVYSkfUirCXs7LCJmjfTxmJmNhI4lqm1pVJoTFpCXXRZbemlmVgs1\nmxJ0wjIzG6c63A/LzMxqoU1W/7XKCcvMbLzylKAN1X1Pzas0/kqTqv/nXva5/gqGlFH1Kr7PvHTr\nSuMDfOf+yyuN39NVpnpBfybNu7/S+AC/ecvHKo2//WXVnrJeuPSKlcYHYMryww7RQlHbtuKEZWY2\nTnmVoJmZ1YNHWGZmVgtedGFmZnXgZe1mZlYPnhI0M7NaqFnCausJTElvz3X+akPSUpL2Hu3jMDMb\nTMcSE1q+tYO2HmFFxGWjfQxDMBXYGzh9tA/EzGxAHW09ZvkPbZ2wJO0BvB1YE/gX8HLglxFxxACv\n2Q04kNQb6w5gH+DVwPeABaRR5a7AP4FvA5sCE4Ejc8v7bwKNq0fPjYgTcvfi8yLiMklvB6ZHxB6S\n7gCuB0TqvfUB4AhgPUn/DVwBfJPUCflZYOch9N4yM6tGzRJWXY52dWAPYBPgzZLe0NeTJK0IHAW8\nOSK2Bp4A9gW2B24E3gocCSwH7ASsFBGbkpo6bizpXaTktjkpae0qaf0BjmsN4EsRsQXw0nx8xwC3\nRcRX8j4uALYBTgZWGOLPb2ZWXE9HZ8u3dtAeRzG4myPisYhYCPyONKLpyxrAX5tGMdeQWtWfQUpe\nlwGfIo20BMwCiIjHI+JLwLrAtRHRExHzSd2D1+u1j+YmZo9ExD/y/X8AS/V67rGkjsdXADuTRlpm\nZu2ho7P1Wxtoj6MY3LqSJkvqAjYDbuvneXeTpuOWzt9vA/wNeC8pEb0FuBD4AnA7aUSEpOUkXZ63\nbZ23TQC2JE0rPg+skmM2j+766tbZzaLf6+7AWRGxHfBX0vSkmVl76Opq/dYG6pKw5pESze+An0bE\nzX09KSIeIU35/VbSDcBKpKm4PwBfkXQlsB/p3NUlwOOSrgMuB74VET8H7pY0izS6uigi/kRaQHGQ\npN8Aqw1yrA8DEyX9L2ka8nRJVwBvBs4Z8m/AzKywuk0JdvT09DVIaB+SVicteNh8tI+lavHwU5X+\nY4xMtfaHK43f01ntz+Bq7YPrnPd0pfHB1dpbMeFlr+4Y/FkDW3D/7S2/5yyx2rrD3t9wtfUqwf5I\neg9wcB8PnRARF4/08ZiZ1VKbjJxa1fYJKyLuIa3aa952CWlKz8zMhsoJy8zM6qBdzk21ygnLzGy8\nqlktQScsM7PxqmYjrLZfJTiePPH0s5X+Y3R1Vr/IZ8lnH600fs+Sy1Qav2vOQ5XGB/jUam+rNP53\n/35RpfEfXnHdSuMDvGT+45XG73z+yUrj90yYXGl8KLNKcN4j/2z5PWfiSi/3KkEzMxsl7jhsZma1\nULMpQScsM7PxygnLzMzqoOrKMaXV62jNzKwcj7Dqr7l+oaTzgI9GxLw+nndWft5lTdsOBa6MiBtH\n6njNzIako8zCP0mdwHeB15Oa5+4dEXc2Pf5u4L9JrZ3OjIjThrIfJ6xBRMT0xXz+16o6FjOzosqN\nsHYCloqILSRtTuq0/l54oVXT8aR2Ts8A10u6JCIW+xqSWiUsScuSWn0sT2qMeBKwCzAbmEZqrrhL\nvn8EqTfVVGBGRJw0QNwvkn7hS5DakVze9Ng9Od4r8r4nktrdT296zmbAicAHga8A5+X97gRMIbU5\n+UpE/HhYvwAzs4IKlmbamtQgl4i4QdLGTY+tC9wZEY8D5JZObyK1jFos9ZrAhLVIU3A7ADuwqGL7\nzIjYFjgfODxvWw14D6lw7kGSVu4roKQNgXeQGkNuCqzDi7sKN3wD+GpEbAGcAGyYt28JHAe8OyLu\n6/WapYHt87EeJ6lWHxDMbGzr6exq+TaIZYHmq7EXNr3f9X5sDrDcUI63bm+gDwEHSno/8BQwIW+/\nMn+dSR6GkpLYXABJtwJrkpor9ibgxohYCCwEDsnnsPp63ix4oVo8knYlJaMpwPw+XnN1RHQDD0l6\nHHgp8K+Wf1ozswoVLHT0FOl9sKEzIhb089gU4Imh7KRuCesQYFZEnCxpO+CdeftGwD+BrUit6AE2\nkNQFLAm8htTqvi+zgf3zScMu4FLgU30873bSHOxvJO0GvCRv/zLwctIJxw/3es1GAJJeRvqUUW13\nQzOzxdBdLmNdD7wbuCCfw7ql6bHbgbUlvQR4mjQd+I2h7KRuCetnwLclTSdl6AWkhLSHpINJJ/Q+\nAqxPGn39ElgRODoiHukrYETcJOky0i+8k3QOa24fT/0ccGo+3/UssDs5IUXE6ZI+mEdczaZKuoI0\n/P1EHsWZmbWFgsVLLwa2lzSTdEplz/x+uExEzMjvz5eT3mPPjIj7h7KT2he/lXQVsF9EzG7atm3e\ntlgr/Aof1x7AtIg4tNXXuPjt4Fz8dnAufjs4F79NnnzmuZbfc5ZbepKL344USfsAvUdAAIdFxKyR\nPh4zs9FWtwFL7UdYY4lHWIPzCGtwHmENziOs5NE5rb/nrDhlskdYZmY2OrprNl5xwjIzG6fqNsPm\nhGVmNk51j/YBLCYnrDZy2yPPVRp/2oqTKo0PMGl+tT/DwonVnhvo6ZpYaXyo/hzTJ9bcudL4xz07\ne/AnDVdPtf8O3XffMviThmHeRu8d/EnDNGHwpwyqZgMsJywzs/FqYc1OYjlhmZmNU54SNDOzWvCU\noJmZ1ULBWoIjwgnLzGycqle6csIyMxu3arbmwglrMI3iuqRGkI81emH1es62jHKxXTOzxbXQU4Jj\nU0ScNdrHYGZWUs3y1egkLEnLAqcDywOrAicBu5CaKU4j9VPZJd8/grT6ciowIyJOGiDuIcB0Up+s\nayLiC5K2Ar5J6gj8LLBzRMzp5/WbAd8i9Wy5H9it6bEvAw8CpwLfBjYFJgJHkts/S5oM/Bj4AfAr\n4PwcaynSCOym1n9LZmbVqtuUYOco7Xct4LyI2IHUYv7gvH1mRGxLeqM/PG9bDXgPaUruIEkr9xVQ\n0vrAh4At821tSe8CdgIuALYhNWdcYYDjOhXYKyI2A34B9FWWeidgpYjYFNgO2DhvX4bUYPLkiPgh\nKaE9CrwD+CSw9AD7NTMbcT09rd/awWglrIeAnST9APgii6qMXJm/zgTUuB8RcyPiOeBWYM1+Yk4D\nboiI+RHRA1wLvAY4ljSKuwLYmTTS6s/UiLgdICLOiIg/9fEcAbPycx6PiC/l7dsAk0gdkCF1O74e\n+CnwFep3jZ6ZjXHd9LR8awejlbAOAWZFxO7AhaQpQMgt54GtgL/m+xtI6srTba8B7ugn5mxgM0lL\nSOoA3gT8jdTK/qyI2C7H3GeA43pA0toAkr4g6X19POd2YJP8nOUkXZ63/wJ4H3CMpFWBbYF/5VHk\n0aTEaWbWNuo2whqtRRc/A74taTrwBOmc05LAHpIOBp4BPgKsTxp9/RJYETg6Ih7pK2BE3CLpAtKo\nphO4DvgJaWrudEnPkEY5AyWsfYEzJXUD/yKdzzqg13MuAd4q6TrS7++opmN4SNKRwPdI3Y3Pk7R/\nft5XWvnFmJmNlPkL2yQTtahtOg43lo9HxOymbdsyjpaLz7zn0Ur/MUaiWvuyT99fafyFS69YafzO\n56rtRAvQ9Uy1XZnHQrX2JeY+Ve0Obrly8OcMw0hUa19m8qRhdwD+/X2Pt/yes8krV3DH4cUlaR/S\n6KW3wyJiVguvfyVwTh8PXR0RRw73+MzM6sLXYQ1RXh3Ye9tVwFW9ts0AZgxjP/eRzi+ZmY1rriVo\nZma1sLBma5edsMzMxqn53fXKWE5YZmbjVN0qXThhtZE5cxdWGn8k/ja7J1S7EnH+EtXGnzSv2lWO\nAA+v2FcBlXKqXsV38ORplcYHOPHf11Uav+sV1f4MC2qSCRbW5DgbnLDMzMYpL7owM7NaqNl1w05Y\nZmbjlUdYZmZWC/Nrtq7dCcvMbJzylKCZmdVC3aYER6u9yGKRdIOk1RfzNf9X0eEg6SxJb8+3Pqu/\nS1pd0g1VHYOZ2XB1d/e0fGsHY3aEFRHvH4F9XFb1PszMqjLupgQlLQucDixP6ux7ErALqaHiNFJz\nxl3y/SNIPammAjMi4qQB4h4DvB34B7BS3rYccAapNxbAZ4BXA++LiD3zc/6UX/eXiJgqaTNSX6tO\n4H5gN2At4MR8bI8Ce0VEn30lckPH04GJwLPA9KbH9gCmRcShkr4I7ET6nZ4MXJ6f0wWcBfw1Ir42\n0O/SzGwkjccpwbWA83Jn3R2Ag/P2mbkC+/nA4XnbasB7gM2BgySt3FdASRuTOgZvAnwUmJIfOhy4\nIncP3oeUGH4BbCFpaUmbAHdFxMNN4U4lJaTN8nPXBU4DPpmP71Lg8wP8fN8AvhoRWwAnABv2cbwb\nAu8ANiM1jFyHlAyXAH5I6q7sZGVmbWX+wu6Wb+2gxJTgQ8CBkt4PPEXqEAzQ6JA2E2h0M5sZEXMB\nJN0KrAk0J5eGdYA/REQ38JSkW/L29YE3S9olf/+SiFgo6SLg/cAWpGTUbGpE3A4QEWfkfa8LfFcS\n+XjvGODnEzArv/6S/Pre/bgE3BgRC4GFwCH5nNvr8+9kmQHim5mNiiqnBCVNAn4ArAzMAT4WEf/u\n43mdpMHETyPilIFilhhhHUIaQewOXEgaWQBslL9uBfw1399AUpekycBr6D9R3AZsKqlT0tLAenn7\nbOD4PDL6EOmXAWma8COkEc6ve8V6IE/rIekLkt4HBPDRHOfzwM8H+PluJ430kLSbpE/38ZzZwBvy\n8U6Q9GtgSeCPwDuBj0h63QD7MDMbcd09PS3fhmB/4JaIeCOpae4X+3ne0cAKrQQskbB+BnxS0tXA\ngcAC0pv1HnnbO4Fj8nMnAL8ErgWOjohH+goYETfl5/0eOI9Fo7BjgA9Jugq4DLg1P//u/PhP86is\n2b7AmflYNiRNAe4PnCPpOuBrwF8G+Pk+BxyW97kbaYqvr+O9DLgeuC4/Z25+7Lmm/S05wH7MzEbU\nwp6elm9DsDXpfRHS+/lbez9B0s6kdQ0tLWDr6KngpFt+c98vImY3bds2b5ve3+vGu8vj4UrPgG68\navUzk1PmPlZp/HmTVxz8ScMw6dE7K40P8O9l16g0/rITuyqNPyaqtT/d52flYuastE6l8QGWX2Zy\nx+DPGtgJ19/V8nvOAVut0e/+JH0cOKjX5oeAT0XE7Xna776IeHnTa14LfAXYGfhv4MHBpgRHdVl7\nvoap9/kggMMiYtYIHsdE4Fd9PBQRse9IHYeZ2Uiat6DMYoq8PuCM5m35WtjGgrkpwBO9XvZR0kK8\nK4HVgXmS7hnocqFKElY+N9R721XAVb22zQBmVHEMiyMi5gHbjvZxmJmNpIr7YV0P7AjcSFpFfW3z\ngxHxwupsSV8mjbAGnBocsxcOm5nZwCpOWCcDZ+e1AvPIs2mSDgbubKy6XhxOWGZm41SVCSsingU+\n2Mf24/rY9uVWYjphmZmNUxWPsIpzwmojr3vZ0pXGX6Jz2IuKBtUzcXKl8Sc++2il8X/zlo9VGh9g\nu5mLPROyeHomVhq+6hV8AJ956daVxv/2g7+tNP7kBU9XGj/vZdgRnLDMzKwWSq0SHClOWGZm49QC\nj7DMzKwOPCVoZma14IRlZma1MMQagaPGCcvMbJyq2wirRLX2ESNpD0lfkzRV0ncHeN49kpYa4j5W\nl3TD0I/yhRpaZmZtbd6C7pZv7aCWI6yIeBD4xGgfR38i4v2jfQxmZoNZ2N0eiahVbZ2wcsfK7wGv\nAiYCF+XtqwPnRcTmkt4FHElqHPknYL+m1+8H7AB8mFS+fjvSz/zjiPjfAXb9UkmXAC8jNXc8Bvgb\nsGlEPCZpf2BKRHy9n+N+MCKm5jYrs4Fp+fh2ycnWzGzUeUqwrP2AeyJiC2A68Fzzg5KWAL4DvDMi\nNgbuBBr9Vj4NvBH4YETMJTVf3DVv613mvrdlSB2MtyRVGV6f1JSx0ctrd+DsFn+Gmbl6/fnA4S2+\nxsyscgu7e1q+tYN2T1gCZgFExB38Z6JZCXg8Ih7Oz/l6RNyXH3srsHxELMzf70bqLnw5sPwg+705\nIp7Mr70RWAc4k9Tq/rXAQxHxUIs/w5X568z885iZtYUF3T0t39pBuyes24FNACStARzb6/GHgeUl\nvSQ/50RJm+bH3gs8Lmm/3Jr+g6Spwe2APSS9aoD9ritpmTyC2wz4a0TcS0qYR9CrUdkgNspftwL+\nuhivMzOrlBddlHUqcKakq4Eu4DjSqAqAiOiW9AngF5IWAn8Gft/0+s+QRkhXAI8BN5CmFX8F3Ef/\nHiNN4b0UOD8ibsvbTwNOJE0JtmqP3P/lGdI0o5lZW2iXqb5WtXXCiojnyU2/+rB5fs4vgV/2emz1\n/PV5YK18/yv5Ntg+7wFe38/DXcCZTdOM/cWY2vTtYRExe7D9mpmNNCesmpC0D30nw8MiYlYfzz+W\nNJ34rvz9e4CD+3j9CRFxccljNTOrghNWTUTEDGDGYjz/8F7fXwIM2Ngorw40M2tLPU5YZmZWB91O\nWGZmVgfdC9tj9V+rnLDMzMYpj7BsyK6978lK42+22rKVxgd4+ZyBrhYYvgUvXbPS+Ntf1vJpzSHr\nfr7af+fuu2+pNH7XK6ZVGh/g2w/+ttL4n566XaXxj3+uHguDe+o1wHLCMjMbr3rcD8vMzOrAU4Jm\nZlYL3W1ScqlVTlhmZuNUt6cEzcysDup24XC7V2tvSa7I/uXRPo6BSNpH0oTRPg4zs4ae7p6Wb+1g\nTCSsmjicVDzXzKwtdHf3tHxrB4NOCUpaFjid1PRwVeAkYBd6tX7P948AuoGpwIyIOGmAuIeQOvgu\nAK6JiC9I2gr4JjAfeBbYOSLm9PP6rYETgMdzjBvy9k+Titr2AOcBJ5P6ar0+Ip6R9FlgYUQc30/c\njwP7k5LLJRFxpKTdgAOBucAdwD6khpDTIuJQSUsBsyNidUlXATcBrwWWJfXhemv+nZwH7NTf78TM\nbCTVbVl7KyOstYDzImIHYAcWVSjvq/X7asB7SK0/DpK0cl8BJa0PfIjUgn5LYG1J7yK9mV8AbENK\nNCsMcFwnAx+OiLcCd+e465GS59bAG3O8NYAfAx/Ir9sVOKef41oZODS/9g3AkrnR41HAmyNia1IT\nx30HOC6AG/Nx/Tof4xnAg6QEbWbWFhYu6Gn51g5aWXTxEHCgpPcDTwGN8zDNrd/f27gfEXMBJN0K\nrEnqCtzbNOCGiJifn3st8BpSR+EjSA0X7wd+N8BxvSwi/pbvX09KrK8FXpVfDynhrU0aIZ4saTYQ\nEfFoPzHXAG6NiOfy94dK2oTUcbgx0ruGlLibj62jV5w/56//II2szMzaTrucm2pVKyOsQ4BZEbE7\ncCGL3pz7av2+gaQuSZNJCeiOfmLOBjaTtISkDuBNwN9InXzPiojtcsx9Bjiu+yWtm+9vkr9Gft12\nefR3FvBOarEuAAAdYElEQVSXiLgjH/fnSF2D+/N3YJqkJQEkXURKuOtJWjo/Z5t8rM8Dq+Rtb+gV\np6+/gm58ztDM2shYXHTxM+CTuU39gaTzRUuSWr9fDbwTOCY/dwKp+++1wNER8UhfASPiFtLU3/Wk\nFvb3AD/J90+XdAXwZvqZusv2Bc7Jz31VjnszaXR1naQ/kEZX9+fnnwFsCPRbpCwi/g38L3C1pFnA\nnyLiXuBI4LeSbgBWIk1HXgasLuk60vTmUwMcK6TfyaU5QZuZjbrunp6Wb+2gYygn3fLCgv2aW79L\n2jZv83maIbrgLw9U+lcxMsVv76w0ftXFbyc8FJXGB+ieOKna+GOg+G33pOUqjT8Wit9OWmqpYX/4\nnXbAT1t+z5l9wntH/cN2pRcOL24b+j5e/0r6HmVdHRFHDuO43N7ezMa9Kqf6JE0CfgCsDMwBPpZn\nsZqfcwgpR3QDxw72/jukhNVX6/eIuAq4qte2xWpD30fM+4D/2NdwtdLe3sxsrFtYbQPH/YFbIuLL\nkqYDXwQOaDwoafn8/VrA0qTLgQZMWF4EYGY2TlW86GJr0rl+SGsb3trr8WeAe0nJamnSKGtAriVo\nZjZOlapgkQsuHNRr80NAo1vpHKCvE5P/AG4jFWr46mD7ccIyMxuneroXFomTiyOc0bxN0v8BU/K3\nU0hFF5q9g3Rp0Kvz95dLuj4ibuxvP05YbWSnZwZdhzIscyftWGl8gPmT1q40fsfC+ZXGX7j0ipXG\nHwnzNnrv4E8ahgUjcE3O5AVPVxq/6lV8B02qfiXlKT33DDtGqYTVj+uBHUmXK72DdGlPs8eB54C5\nEdEj6QlSCcB+OWGZmY1T3fPnVRn+ZODsfK3qPPKKcUkHA3dGxCWS3grcIKkbuI5Uzq5fTlhmZuNU\nlSOsiHiWVPy79/bjmu4fSSrM0BInLDOzcariKcHinLDMzMYpJywzM6sFJywzM6sFJyx7gaQvAw9G\nxCmjfSxmZr0tXFDpKsHinLDMzMYpj7BGkaRlSd2FlwdWBU4CdiE1jJxGauK4S75/BKl21VRgRkSc\n1E/M1Uk9wR4FLiV1Gj6SVIdxGdK1BfOAH5HKjKwJ3BgR+zfFWAs4F9g7Iv5S8mc2MxuqnoX1Slhj\nrfjtWsB5EbEDqY19o4XIzFxh/nzg8LxtNeA9wObAQZJWHiDuVGCHiPg6qZPy7jne/7HoOoN1gI8D\nmwI7Spqat4uUrHZzsjKzdtLTvbDlWzsYUyMsUrHFAyW9n9QBeELefmX+OhNo1K2ZGRFzASTdShoZ\nPdxP3LsjojHZez9woqSnSUnv+rz9zoiYk+P9C1gqb38HqUtze/yLm5ll7ZKIWjXWRliHALMiYnfg\nQtIUIMBG+etWwF/z/Q0kdUmaTBo13TFA3Oay96cBe0bEHsADTfvor8Dat0hVjM+W1LUYP4uZWaXq\nNsIaawnrZ8AnJV0NHEga2SwJ7JG3vRM4Jj93AqlHy7XA0RHxSIv7+AFwraTrSRWIVx3sBRHxa1IJ\n/S8sxs9iZlap7gXzW761g46enuorL48mSVcB+0XE7KZt2+Zt00fruPoyb9aPK/3HmPv66qu1Txi8\nB9uwVF2tvfPZxyuNPxLmTpk6+JOGYSxUa1+41LKVxh+hau0dgz9rYFPe9NmW/zHnXPONYe9vuMba\nOawhk7QPuZpwL4dFRLV9P8zMRkF3m0z1tWrMJ6y8mq/3tquAq3ptmwHMGJGDMjNrA3Vb1j7mE5aZ\nmfWtXRZTtMoJy8xsnOp2aSYzM6uDuo2wxvwqQTMzGxvG2nVYZmY2RjlhmZlZLThhmZlZLThhmZlZ\nLThhmZlZLThhmZlZLThhmZlZLThhmZlZLbjSRQ1JWonUIfmOiHison10kppTbgn8rqnjctuTtF1E\n/Ha0j2O48r/BS4GHI6L2V/hLmhAR7dFYyWrJCatmJH2C1MH4VmA9Sf8TET8ovI9vAbcDrwLeADwE\nfKxA3H36eyxXyy/lKKDShCVpaWAFYD6wD3BORNxbMP77geOAx4EpkvbPjUBLxX8dsDSpm/axwLER\ncUWp+Hkfn4+Ir+f76wNnk/6eSsVfGTgCWIfUSfyYiCjS0Cx3B+8CzgN2IX146wQujYg3l9hH3s9r\nI+LWfL8D+EJEfK1U/LHGCat+/gtYPyKelzQZuJrUBbmkTSLiQEm/jYjtJJV6I1ulUJzB9Ei6GAjS\nGzIRcXjhfVwEnAJ8gNRNegbwtoLxvwRsGhEPS3oZqZt2sYRFOvZPkZL7EcDXgaIJC3itpP2AZYCP\nAvsXjn8+cAFwJrAV8H3gXYVi7wUcDkwl/R11kP6Wri0Uv+EMSR/Osc8hJV7rhxNW/TwELMj3nwMe\nrWAfXZI2Au6RNBGYUiJoRBzVuC/prcAawA3A30rEb3Jm4Xh9mQxcAhwQER/NP09Jj0bEwwAR8ZCk\npwrHf5705jgxIm6QVEUV1D2AH5KmNTeJiLmldxARJ+e7N0v6UMG4pwGnSdorIqr8e9oV+BEwCTio\n9Ch3rHHCqp9O4CZJM4ENgQmSzgWIiL46Jg/F2aRPrnsAx5M+jRcj6Vjg5cC6wFzgMODDBXdxd8FY\n/ZkIHAD8UdJ6pOm1kuZIupw0gt4YmJx/b6VGiz2kT/SX5jf6YueWJM3K8QEmAK8HfiuJiNiy1H6A\n2ZJ2I03/bgQ8KmkdgIgo9SHo15I+DyzV2BARXxlu0F7T49cD7wDWlLRm4enxMcUJq36OyV97SJ9e\nq/AosBA4BFifNAoqaeuIeFOecjxbUumpoka8DuA1wD3ANYX3cQiwE+nfY3fgM4Xj/6Tp/v2FY0M6\nL7Mp8Etg2/x9KdOb7neQ/laXJH04KWlavu3dtO3UvL9S55kuBH4D/KNQvIbm6fEnSefKRmrKvLac\nsOrndio60dzkIOANEfG0pCnAlaTzA6UsIWkp0rmmLlJyLCYiXhit5SnNC0rGz94cEZ/P978j6avA\n74cbVNLGEfEH4F+9H4uIXw03fpPtI+JHeZ+zgXNJn/KHrbH4RNJ/AetExOck/Yr0N1Ts7ygitisV\nawBzIuKLpYP2mh5fmaYRnPXPCat+zs+3Kk40N3RHxNMAETFH0vOF4x8P/JF0buN3+fuqLEE6V1aE\npI+TPtGvK2nHvLmTNEV4WIFdvAX4A/85RdoDlExYH5E0h/RGeQxwZMHYDfuTRnEA7ySNcoedsCRd\nFBE7S/oXi6YeAYiIVYcbv5dbJU0H/tzYV8HpRiSdRPrdPMCi0WjJadMxxQmrhiKicU6p6InmJndJ\n+ibpDeZNwN9LBo+ICyX9BlgLuDsiHikZv+mNrIP0N/6tguF/QFpNdziLpme7gYdLBI+I/813vxQR\n/2xsz2+aJX2AtPJwKdIU7b8LxwdYGBELACJivqQi15JFxM75a59TaJLeGxE/LbEvYIN8ayg53Qiw\nGbBGRHQXjDlmOWHVz2xJu5Om6ao60bwnsC+wPWkK8tBCcQGQtAHp2qWl8vdExF6l4vf3RlYo9lzS\n6sn9SIshGlM5r6bsebILJb2LtCL0ZNI1X+cNN6ikH7FoVPIcaQR0Qv43KLVop+Gnkq4FbiRdf3VJ\n4fj9OQAokrBGYNrxTtLf0LMV72dMcMKqn2mASNeJQDqZXfREc/5UfFKJWP04C/gO5U9kAy9cpHom\naSXig8BeEfHnwru5CFiZRT9DD2UTVuNNdzng+IJLq3uv+PxGobj/ISKOlvRz0t/rORFxc1X76qWj\nVCBJd/PiaccnI2LDUvGBVwL3Srozf99TeCXlmOKEVT/nAweTlgt3APMjYu3RPaTF9mBEnF5h/BOB\nvSPi5jyaO4l0vq+kqVW8sUjaoenbK0ij3H9K2qHEoouIuDrv593AxhFxpKTLKHgeUdLeEXF6XojS\neLN/vaRdKriAuy8ly1hNy187SDMaHywYG8pezjHmOWHVzyeAbYAvkpbcHjC6hzMk90g6lBefyC65\noKCj8Wk+Im6StGCwFwzBbEmrRsQDheP2fgMLFi0TL/k7OgpoTHftQlrefnmh2I1R5+xC8UZNr4ud\nr89JuKQJpCTY+AC6Kmk63vrghFU/D0TEvyRNiYirJFWxuqtqS5KmiZS/L70CbmE+/3MtadFI8QoL\nwBuB+yQ1Fiv0lFihFhF7wgsFjjeMiF9L+hTly2/Nj4gn8z6fLFzpoiePFP9jaf4IKTkl2DxKXJVc\n6qugc4GLga1JKwWXKRx/THHCqp8nJe1EelPYF1hptA9ocTXelCu0F+nczNdIdf7+q/QORmAa9kfA\nCfn+Y6SEVfLyhRtzhZRZpIUXJc/xNY8SG6s1G2/6xT6YSPoscHYfKxyPK7UPXjxKvBm4rGBsgKcj\n4quS1o6IvfIiFeuHE1b97E1aDn4YqdrCp0f3cBafpMOAL5BWRnVQaHTSEBH35uoZS1P2fMYLJH2v\nd+ySKx2BpSPi5znuufki3GIi4tP5g4+ACyLiZwVjj9Qo8WngYkkPAmcAl0VET8mfhVRNZl9gPVLN\ny+cKxob0wXMqqSL/0niENSAnrJqJiDks+jR8yGgeyzBMB1aNiEqW8kqaQVox+TDVXYzZWGLeQVqy\nXfqC1XmStieVxdqUQtVAJL0rIn7eVMvucWAVSftUUMOu0lFivh7xFEmvIVV/OVXSmcAJBau/nAo8\nQaqUvw1wOqnyfClHkUp8fR+4i7IVZcYcJywbDXdT/pNqs9cBa1fZ9DAimhcoXJZLD5W0N2la80TS\ntGapE/Er5q+9r1Wr4ndV6ShR0vKkDz8fJSWVA0g9rH5OuVWha0fEm/L9n+Si0yVtGhGNSwtG6jq1\n2nLCstEwEbhF0i35+57CF60+QGqJUrolxwt6LT9fBXhZobhL5Ovg7gM+xIvP/5QwK19o/qOCMftT\nySixye9Jo7bpEXFfY6OkktdJLSVpckQ8K2kSKSGWtKOk4yOiivYuY44Tlo2G/x38KYuvqa3FysAd\nku7KD1VxMWbzwoLnWXQh93CdQ+qRFCxKVI2kVaIm4qlN93snwpIlh6C6UWLDOsBrgU1yUpkNEBFH\nFNzHCaQSaLeSzmOVXpW7EvBA0wXKvnB4AB09PZXNmpj1SdJLSN15X7j2JCKGfX2LpFf191heiLFZ\nRPxuuPtp2t9rySfjI+KmUnFHQq6Wv25E/DkvvvhFRBTriTXIvk+OiGG3lJF0BKnC/O9JNfkuiIiS\ndSMb+3kl6UPQw80juUKx/+Nvtoq/1bHCIywbDReTahSuTxqdFFl80WhrMYCvUmgUIenTpJHQ74DP\nSrqg6VzEcOI2Nz98kcKfvH8A/IK0gGcd0vRj6VqC/dHgT2nJu4CtIqJb0hLAdZQtdEy+znHJiDhc\n0oWS/tBUoHjYBvibLfa3OpZ0jvYB2LjUERH7kaa9tgdeMlL7LRhrV+CNEXEg6QR/qQaI00nTjR/O\n+/gwqfNz6RI+q0XE9wAi4uvUs3ngQ8DkfH8iUEXF+fc0yklFxAeB91Swj76U/FsdM5ywbDQsyFNS\njeukRmqkX3L+u6O5dQaFWsxHxL35U/cOwCfz/e+QKnaU1NOo8i9pTcovJqiMpFl5td5rSecqLyNd\n4PvyCnbXnZuAImkCI/ee6XM1ffCUoI2Gk0hdjX9Fqjt33egezpBcJ+kiUvmnNwLXF45fSfPDJgcC\n50t6GWlVZZ3q1w3YG6zw+Z9TSE0cbyEVwq1kwZC1xgnLRlxE/LhxX9KFEVHZ8vNeik2zRMRnJb0T\nWBc4MyIuLRU7q6T5YUNE3Aj8x/JvSUc2t2+vyLD+HUbyXGVEnCHpEtIKzb83mo0WbhLZF08J9sEJ\ny0acpL1JI6xJ+Xsiolgb+xxzWWB10pvMM3nzuQXjv6g9h6QFhSvO925+WOWbY7NtSgWSNIVUgmtV\n0sW8f4mIO0nTnVUq+mafaxX2Pj9WrElkLmG1JnBHRDyWNxf7Wx1LfA7LRsP+wI6k0UnjVoyknYGr\nSXXgDpb0RYCIOK3gbo5iUZHVXYAvF4xNRBxNqhN5I3BgY2WapM1K7qcPJd/szySVG1qb1EjzDHjh\nnF+VRuL8T5Hfk6RPkAoQH0q6qHt3KP63OmY4YdloeCQvLpjbuBWOfxCwOfAIcDTwvsLxoVd7DspX\ncSAiboqI83t16i3dj6m3km/2K+ZOyfMjYiZj6/2m1O/pv4D1I+J9pCnaOva3GzGeErQRI+nYfHei\npMuBP7GogWPJTrQLI2KupJ6I6JH0zOAvWWxVtucYSK3ObUialr++HKiikWZf6vQ7eohFv5fngEdH\n8Vja3lj6xGPtL/LtcuBS0lLkj5FGQiVdl5PJyyWdQqqEUFREfBq4gHQd0AUR8ZnS++hH1dNdJd/s\nPwN8j3QO7iIq7C4g6RVN347E+Z9Sv6dO4KbcYeBGYKqkc/Pfr/XiEZaNmIg4G0DS70kFS/+eFxac\nRcGme7kqwdtJo57ZhfsjASBpNVJ/pNuBz0v6R53KM+XeSyuQrh/bBzgnr74r2TrjDuATTeWfbhns\nBYtD0udIVdqXB/aUdFlEHFzy/I+kA0i/m97tSkr9vR6Tv/aQzrnaADzCstEwPyL+DhARd1Go7bik\nLkkTJf0fcAWpcOlvJF1ZIn4v55IqtB9D6pV0fAX76EupT/YXARsB/4+UtGYARMQ/CsWH9Aa8Qb6/\nDnB2wdgAH8gx3xER6zXtq6QlSH9DP5S0bWNjwQ9BtwPvJzVkfTdpJeXVEXF1ofhjihOWjYZ7JR0r\n6d2S/ge4v1DcvUhTju9g0fTjLaRWHaV1ky7mXT4izqNQ0m0maSVJm+ViwQ2lpoomk/ovvTwivkY1\nlS6qLv+0EJhKOg8Ei8o0FRMR34yIjUg1Cj8h6W+Fd3E+KWkdihs4DspTgjYa9gT2Iy1tv520km/Y\n8lTQaZL2yqvTqjQB+DpwjaTtSLXsisnLnQ8CbgXWk/Q/EfGDgtNdE0kr0v4oaT1SmazSeiStExF/\nq6j801X5truk40nFfIvKPbA+QDrX2kH59iKNzsmQ2ph8qHT8scQJy0ZcRDxP4aravVwj6TBe3L6k\ndOmhPUmFe08ntTj/WOH4jeXOz0uaTLqu7AcF4x9COu6jgY9QzXLqg0jln6aSRtH7FY7/5UbvK0l/\nAF4xyPOH4i+k6dP980XPpc3O115dSZqifbRR4zEiSo/mas8Jy8aic0ktTLYm1clbpoJ93EVKhseT\nFl/8s3D8Spc7R8RMSSuQFlz8LZdqKirX8yvZ/be3HwE75/t7kpLwOoX3sS6pLNM6kp4H7o+Ikis1\np5HarTQagC5JarLZg9uL/AcnLBuLno6Ir0paOyL2yisRS5tBWqH2a1I5o9Mpu8Kusdx5JulNf0Jj\nqXNEDLtvlaSvkipQXAd8TNKbIqLIsnNJF0XEzpL+Ra9l+BGxaol9ZL+R9H3SKsEnSE0cS9uPdOH5\nS0gLPNYCPlUw/vnAwSyaDZgfEWsXjD+mOGHZWNSTp6Gm5OXbVYyw1o6IRsuPn+TEUlLVy53fFBFb\nAUg6AbihVOCIaIx6Xpfr8BXVaPdBKv20DPAW4OOl95NNJ7V2uSIivpUvySjpE6QPPF8ELsSVLgbk\nVYI2Fh1FOj/zfeDvpCXupS2Vzy01TsyXXlBQ9XLnCZIa//87qOaC5MskXSBp+8Jxg3TReQCfJE2r\nzc630jpJv5vG76d0GbEHIuJfwJSIuIo0WrR+eIRlY4aku1n0xtJBur7oOVI/qc8W3t0JpFVdtwLr\nUX712Pn5diapo/H3SS3hS8a/XtINpKm08wrGBiAiNpK0Memi3mOBn0TEMYO9roW4rwaQ1EFalv8P\nSZtERPGKJqTzodcAr5J0KfCTwvGfzBdV90jaF1ipcPwxxSMsG0umkZLHb4FdImId0iil+DmsiPgh\n6Y3+GGDLfC0Wkt5bcB+nRMTNEfFdCk9rRsQ3SSsRrwf2iYhvQSXV4P9Kqrf4GKnRZUknkyrlQ1ra\nXnzlaUR8h7Qw5RDg0Ij4RuFd7A3cSxpJr0Oq0G/96OjpcSdmG1skXRUR2zZ9f03T+aaq931lRAx7\ndZekk0nJpLHceQ/Sm1qly51LHX+OdSapav5FpCaX95SI2xT/xojYtOn7Yv/Okv67v8ci4isl9mGL\nz1OCNhY9kSto3AhsCfxrBPddqnTSaC13Lln89mLg44WXgb+IpBUj4lFJy1P2/axRPWMn4G7Sh4dN\ngFcW3IctJicsG4t2Iy1HfhdwG4WbKw6i1JvzaC13Lplc7iZdxL0C6aLnWyPi5wXjfwX4g6THgeVI\nCzCKiIhTASR9ICI+kTf/UNKvS+3DFp8Tlo05EfEM8M3RPo5hGgvLnU8gXdB7Gqnb8C+BYgkrIn4u\n6ZekhQoPVzSSe4mkNXNnAZESo40SL7owK6vUlNpoLXcu2vwwlzPqyddjzSkRU9J38tdZpAU1F5NW\nPJa+Fg7gQOBiSQ+SVgwWG8XZ4vMIy2yIJG1AOs90W0Q0ej2V6pNU+XJnSSsBawJ3RMRjeXPJxoGP\n5WNfWtJ0UjWKEv4nf51eKF6/IuI64HW9t0s6MiKOqnr/9mIeYZkNgaSjge+SVsGdmpsJluyTVOly\n51wNfhaprcWsXIC1UfG+lI8DryZ1lN6YctUoDsjXde3bx22kbDOC+7LMIyyzoXk7sGlEdEvqIr35\n/79SwSNiDqljMlTTWr6yavCSmlfSfbfp/jKk67GGK/LXqcDzpJHbsYzsecuiU6fWGicss6G5H5gC\nPElayffQwE9vO1VWgz8/f12R9Du6BXhN3ucbhhs8Is4GyHX9pucFEdcCZ1FuSnYwvoB1FHhK0Gxo\nVgH+JulXwB2kJoszKzrxX4VGNfgZpOvVpko6t1ERfjgiYouI2IJU5WKdiNiBNK1ZugXL/Ij4e97n\nXVTQ9dnai0dYZotB0t4RcTqpqO7f8+brSJ+4Dxu1A1t8VVeDh1Tnbw6kSw0krVI4/r35XNYsYFPS\nqHekeEpwFDhhmS2ef+Svl/V+ICLuHeFjGY7bgSNII5+/AsdExOOF9/ErSVcDfyAllNKFY/ckXSC+\nI+nnObpwfHJ7mhVIhZT3Ac7J/84le59Zi1xL0GwckvRb0rmmWaRq8DtGRMlq8I39bERKirdFxM15\n22a5G3HbyxcmnwJ8gFQ1ZbuIeNvoHtX45RGW2TgVEafkuzdL+lBF+/gj8Mdem79Kfdq/TwYuAQ6I\niI9KeutoH9B45oRlNj7NztdeNarBPyppHai2GnxWp/M/E0llsf4oaT1g6VE+nnHNCctsfBqtavBQ\nryXhh5Aqth8NfIR61nQcM5ywzMan0aoGXysRMTNXm98H+FtE3DjaxzSe+Toss/GpUQ3+UlJzyFtH\ncN+1mRKU9FXSasT5wMck1b0LQK05YZmNTyNSDV7SSpI2k/SSps0lC+xW7U0RsXNEfIu0UnDr0T6g\n8cwJy2x8Golq8CNRYLdqEyQ13ic7qNf5tzHHCctsfKq0GnzWKLD7PmBD6rlg4XxSr63jSRVNzhvl\n4xnXvOjCbBwagWrwUG2B3REREd+UdDlpVeUZEXEr1Ovi57HElS7MrBK5MPCqwEzSCGsCqVoEEbHr\nKB7asEm6MiLqcvHzmOERlplVZSQK7I6W2qx0HEucsMysKiNRYHe0eGpqFHjRhZlV5XxS0joUuAv4\n/ugejtWdR1hmVpmRKLA7SjwlOAqcsMysKqNZYLcYSSsBawJ3RMRjeXOdLn4eM7xK0MwqkXtuNb/B\nLAnMA3rqssIuX/x8EKl01XrA/0TED0b3qMYvj7DMrCpjocBu4+Ln5yVNBq4GnLBGiRddmFlVRrPA\nbim1v/h5LHHCMrOqjEiB3Yp1AjdJmgHcCEyVdK4kn8MaBZ4SNLOqVF5gdwSM5Yufa8cJy8yqsjew\nFqnA7iFUU2C3amP54ufa8SpBM7N+5JWO55PapGwF7BgR7xrdoxq/PMIyMxvAGL74uXacsMzM+jcm\nLn4eK5ywzMz6Nw0QsFf+fkngVNIijFpc/DyWOGGZmfVvLFz8PGb4Oiwzs/6NhYufxwwnLDOz/o2F\ni5/HDCcsM7P+jYWLn8cMJywzs/7tDdxLuvh5Hep58fOY4QuHzcysFjzCMjOzWnDCMjOzWnDCMjOz\nWnDCMjOzWvj/Y6f0Zd+nGXAAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x1af0e700b8>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "sb.heatmap(train.corr())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 83,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x1d7f51eba8>"
      ]
     },
     "execution_count": 83,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAawAAAFDCAYAAABm2jr7AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nWGZmVgv/Hx776nc9HnqnAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x1afff92e48>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "sb.heatmap(test.corr())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 84,
   "metadata": {
    "scrolled": true
   },
   "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>app</th>\n",
       "      <th>os</th>\n",
       "      <th>channel</th>\n",
       "      <th>date_hour</th>\n",
       "      <th>ip_oscounts</th>\n",
       "      <th>app_clickip</th>\n",
       "      <th>app_os_clicks</th>\n",
       "      <th>app_device_click</th>\n",
       "      <th>clicks_by_ip</th>\n",
       "      <th>app_os_dev_count</th>\n",
       "      <th>app_rank</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>count</th>\n",
       "      <td>1.879047e+07</td>\n",
       "      <td>1.879047e+07</td>\n",
       "      <td>1.879047e+07</td>\n",
       "      <td>1.879047e+07</td>\n",
       "      <td>1.879047e+07</td>\n",
       "      <td>1.879047e+07</td>\n",
       "      <td>1.879047e+07</td>\n",
       "      <td>1.879047e+07</td>\n",
       "      <td>1.879047e+07</td>\n",
       "      <td>1.879047e+07</td>\n",
       "      <td>1.879047e+07</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mean</th>\n",
       "      <td>1.221480e+01</td>\n",
       "      <td>1.873312e+01</td>\n",
       "      <td>2.648059e+02</td>\n",
       "      <td>9.219946e+00</td>\n",
       "      <td>6.180820e+01</td>\n",
       "      <td>3.648960e+04</td>\n",
       "      <td>3.321143e+04</td>\n",
       "      <td>3.384699e+04</td>\n",
       "      <td>1.197942e+04</td>\n",
       "      <td>3.443016e+04</td>\n",
       "      <td>2.092161e+01</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>std</th>\n",
       "      <td>1.164924e+01</td>\n",
       "      <td>1.135059e+01</td>\n",
       "      <td>1.355254e+02</td>\n",
       "      <td>3.751926e+00</td>\n",
       "      <td>3.481484e+01</td>\n",
       "      <td>2.365194e+04</td>\n",
       "      <td>1.778136e+04</td>\n",
       "      <td>2.192823e+04</td>\n",
       "      <td>1.400843e+04</td>\n",
       "      <td>1.885363e+04</td>\n",
       "      <td>2.198813e+01</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>min</th>\n",
       "      <td>0.000000e+00</td>\n",
       "      <td>0.000000e+00</td>\n",
       "      <td>0.000000e+00</td>\n",
       "      <td>4.000000e+00</td>\n",
       "      <td>1.000000e+00</td>\n",
       "      <td>1.000000e+00</td>\n",
       "      <td>1.000000e+00</td>\n",
       "      <td>1.000000e+00</td>\n",
       "      <td>1.000000e+00</td>\n",
       "      <td>1.000000e+00</td>\n",
       "      <td>0.000000e+00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25%</th>\n",
       "      <td>3.000000e+00</td>\n",
       "      <td>1.300000e+01</td>\n",
       "      <td>1.350000e+02</td>\n",
       "      <td>5.000000e+00</td>\n",
       "      <td>4.300000e+01</td>\n",
       "      <td>8.542000e+03</td>\n",
       "      <td>1.795400e+04</td>\n",
       "      <td>1.123100e+04</td>\n",
       "      <td>2.300000e+03</td>\n",
       "      <td>1.741800e+04</td>\n",
       "      <td>9.000000e+00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>50%</th>\n",
       "      <td>1.200000e+01</td>\n",
       "      <td>1.800000e+01</td>\n",
       "      <td>2.360000e+02</td>\n",
       "      <td>1.000000e+01</td>\n",
       "      <td>5.500000e+01</td>\n",
       "      <td>4.454300e+04</td>\n",
       "      <td>3.383100e+04</td>\n",
       "      <td>4.438500e+04</td>\n",
       "      <td>6.021000e+03</td>\n",
       "      <td>3.588900e+04</td>\n",
       "      <td>1.200000e+01</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>75%</th>\n",
       "      <td>1.800000e+01</td>\n",
       "      <td>1.900000e+01</td>\n",
       "      <td>4.010000e+02</td>\n",
       "      <td>1.300000e+01</td>\n",
       "      <td>7.200000e+01</td>\n",
       "      <td>6.083400e+04</td>\n",
       "      <td>4.792400e+04</td>\n",
       "      <td>5.348500e+04</td>\n",
       "      <td>1.707500e+04</td>\n",
       "      <td>5.325000e+04</td>\n",
       "      <td>2.000000e+01</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>max</th>\n",
       "      <td>5.210000e+02</td>\n",
       "      <td>6.040000e+02</td>\n",
       "      <td>4.980000e+02</td>\n",
       "      <td>1.500000e+01</td>\n",
       "      <td>1.990000e+02</td>\n",
       "      <td>6.429700e+04</td>\n",
       "      <td>6.553400e+04</td>\n",
       "      <td>6.471400e+04</td>\n",
       "      <td>6.471800e+04</td>\n",
       "      <td>6.536100e+04</td>\n",
       "      <td>2.070000e+02</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                app            os       channel     date_hour   ip_oscounts  \\\n",
       "count  1.879047e+07  1.879047e+07  1.879047e+07  1.879047e+07  1.879047e+07   \n",
       "mean   1.221480e+01  1.873312e+01  2.648059e+02  9.219946e+00  6.180820e+01   \n",
       "std    1.164924e+01  1.135059e+01  1.355254e+02  3.751926e+00  3.481484e+01   \n",
       "min    0.000000e+00  0.000000e+00  0.000000e+00  4.000000e+00  1.000000e+00   \n",
       "25%    3.000000e+00  1.300000e+01  1.350000e+02  5.000000e+00  4.300000e+01   \n",
       "50%    1.200000e+01  1.800000e+01  2.360000e+02  1.000000e+01  5.500000e+01   \n",
       "75%    1.800000e+01  1.900000e+01  4.010000e+02  1.300000e+01  7.200000e+01   \n",
       "max    5.210000e+02  6.040000e+02  4.980000e+02  1.500000e+01  1.990000e+02   \n",
       "\n",
       "        app_clickip  app_os_clicks  app_device_click  clicks_by_ip  \\\n",
       "count  1.879047e+07   1.879047e+07      1.879047e+07  1.879047e+07   \n",
       "mean   3.648960e+04   3.321143e+04      3.384699e+04  1.197942e+04   \n",
       "std    2.365194e+04   1.778136e+04      2.192823e+04  1.400843e+04   \n",
       "min    1.000000e+00   1.000000e+00      1.000000e+00  1.000000e+00   \n",
       "25%    8.542000e+03   1.795400e+04      1.123100e+04  2.300000e+03   \n",
       "50%    4.454300e+04   3.383100e+04      4.438500e+04  6.021000e+03   \n",
       "75%    6.083400e+04   4.792400e+04      5.348500e+04  1.707500e+04   \n",
       "max    6.429700e+04   6.553400e+04      6.471400e+04  6.471800e+04   \n",
       "\n",
       "       app_os_dev_count      app_rank  \n",
       "count      1.879047e+07  1.879047e+07  \n",
       "mean       3.443016e+04  2.092161e+01  \n",
       "std        1.885363e+04  2.198813e+01  \n",
       "min        1.000000e+00  0.000000e+00  \n",
       "25%        1.741800e+04  9.000000e+00  \n",
       "50%        3.588900e+04  1.200000e+01  \n",
       "75%        5.325000e+04  2.000000e+01  \n",
       "max        6.536100e+04  2.070000e+02  "
      ]
     },
     "execution_count": 84,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "test.describe()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 85,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "758"
      ]
     },
     "execution_count": 85,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "gc.collect()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 86,
   "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>app</th>\n",
       "      <th>os</th>\n",
       "      <th>channel</th>\n",
       "      <th>date_hour</th>\n",
       "      <th>ip_oscounts</th>\n",
       "      <th>app_clickip</th>\n",
       "      <th>app_os_clicks</th>\n",
       "      <th>app_device_click</th>\n",
       "      <th>clicks_by_ip</th>\n",
       "      <th>app_os_dev_count</th>\n",
       "      <th>app_rank</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>count</th>\n",
       "      <td>1.452407e+06</td>\n",
       "      <td>1.452407e+06</td>\n",
       "      <td>1.452407e+06</td>\n",
       "      <td>1.452407e+06</td>\n",
       "      <td>1.452407e+06</td>\n",
       "      <td>1.452407e+06</td>\n",
       "      <td>1.452407e+06</td>\n",
       "      <td>1.452407e+06</td>\n",
       "      <td>1.452407e+06</td>\n",
       "      <td>1.452407e+06</td>\n",
       "      <td>1.452407e+06</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mean</th>\n",
       "      <td>1.713381e+01</td>\n",
       "      <td>1.902599e+01</td>\n",
       "      <td>2.491856e+02</td>\n",
       "      <td>9.101441e+00</td>\n",
       "      <td>5.336020e+01</td>\n",
       "      <td>2.885798e+04</td>\n",
       "      <td>3.002873e+04</td>\n",
       "      <td>3.116766e+04</td>\n",
       "      <td>1.024507e+04</td>\n",
       "      <td>2.923434e+04</td>\n",
       "      <td>1.683267e+01</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>std</th>\n",
       "      <td>2.149883e+01</td>\n",
       "      <td>1.155443e+01</td>\n",
       "      <td>1.284466e+02</td>\n",
       "      <td>6.074046e+00</td>\n",
       "      <td>3.688474e+01</td>\n",
       "      <td>2.279913e+04</td>\n",
       "      <td>1.934144e+04</td>\n",
       "      <td>2.169428e+04</td>\n",
       "      <td>1.343970e+04</td>\n",
       "      <td>1.975315e+04</td>\n",
       "      <td>2.028013e+01</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>min</th>\n",
       "      <td>0.000000e+00</td>\n",
       "      <td>0.000000e+00</td>\n",
       "      <td>0.000000e+00</td>\n",
       "      <td>0.000000e+00</td>\n",
       "      <td>1.000000e+00</td>\n",
       "      <td>4.000000e+00</td>\n",
       "      <td>1.000000e+00</td>\n",
       "      <td>1.000000e+00</td>\n",
       "      <td>1.000000e+00</td>\n",
       "      <td>1.000000e+00</td>\n",
       "      <td>0.000000e+00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25%</th>\n",
       "      <td>5.000000e+00</td>\n",
       "      <td>1.300000e+01</td>\n",
       "      <td>1.350000e+02</td>\n",
       "      <td>4.000000e+00</td>\n",
       "      <td>3.300000e+01</td>\n",
       "      <td>3.860000e+03</td>\n",
       "      <td>1.217200e+04</td>\n",
       "      <td>1.235700e+04</td>\n",
       "      <td>1.059000e+03</td>\n",
       "      <td>1.043700e+04</td>\n",
       "      <td>6.000000e+00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>50%</th>\n",
       "      <td>1.200000e+01</td>\n",
       "      <td>1.900000e+01</td>\n",
       "      <td>2.340000e+02</td>\n",
       "      <td>9.000000e+00</td>\n",
       "      <td>5.100000e+01</td>\n",
       "      <td>2.468800e+04</td>\n",
       "      <td>3.128900e+04</td>\n",
       "      <td>2.292900e+04</td>\n",
       "      <td>4.477000e+03</td>\n",
       "      <td>2.658800e+04</td>\n",
       "      <td>1.000000e+01</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>75%</th>\n",
       "      <td>1.900000e+01</td>\n",
       "      <td>2.200000e+01</td>\n",
       "      <td>3.430000e+02</td>\n",
       "      <td>1.300000e+01</td>\n",
       "      <td>6.500000e+01</td>\n",
       "      <td>5.042700e+04</td>\n",
       "      <td>4.586000e+04</td>\n",
       "      <td>4.706800e+04</td>\n",
       "      <td>1.486100e+04</td>\n",
       "      <td>4.755100e+04</td>\n",
       "      <td>1.900000e+01</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>max</th>\n",
       "      <td>5.020000e+02</td>\n",
       "      <td>5.660000e+02</td>\n",
       "      <td>4.980000e+02</td>\n",
       "      <td>2.300000e+01</td>\n",
       "      <td>1.990000e+02</td>\n",
       "      <td>6.429700e+04</td>\n",
       "      <td>6.553400e+04</td>\n",
       "      <td>6.471400e+04</td>\n",
       "      <td>6.471800e+04</td>\n",
       "      <td>6.536100e+04</td>\n",
       "      <td>2.070000e+02</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                app            os       channel     date_hour   ip_oscounts  \\\n",
       "count  1.452407e+06  1.452407e+06  1.452407e+06  1.452407e+06  1.452407e+06   \n",
       "mean   1.713381e+01  1.902599e+01  2.491856e+02  9.101441e+00  5.336020e+01   \n",
       "std    2.149883e+01  1.155443e+01  1.284466e+02  6.074046e+00  3.688474e+01   \n",
       "min    0.000000e+00  0.000000e+00  0.000000e+00  0.000000e+00  1.000000e+00   \n",
       "25%    5.000000e+00  1.300000e+01  1.350000e+02  4.000000e+00  3.300000e+01   \n",
       "50%    1.200000e+01  1.900000e+01  2.340000e+02  9.000000e+00  5.100000e+01   \n",
       "75%    1.900000e+01  2.200000e+01  3.430000e+02  1.300000e+01  6.500000e+01   \n",
       "max    5.020000e+02  5.660000e+02  4.980000e+02  2.300000e+01  1.990000e+02   \n",
       "\n",
       "        app_clickip  app_os_clicks  app_device_click  clicks_by_ip  \\\n",
       "count  1.452407e+06   1.452407e+06      1.452407e+06  1.452407e+06   \n",
       "mean   2.885798e+04   3.002873e+04      3.116766e+04  1.024507e+04   \n",
       "std    2.279913e+04   1.934144e+04      2.169428e+04  1.343970e+04   \n",
       "min    4.000000e+00   1.000000e+00      1.000000e+00  1.000000e+00   \n",
       "25%    3.860000e+03   1.217200e+04      1.235700e+04  1.059000e+03   \n",
       "50%    2.468800e+04   3.128900e+04      2.292900e+04  4.477000e+03   \n",
       "75%    5.042700e+04   4.586000e+04      4.706800e+04  1.486100e+04   \n",
       "max    6.429700e+04   6.553400e+04      6.471400e+04  6.471800e+04   \n",
       "\n",
       "       app_os_dev_count      app_rank  \n",
       "count      1.452407e+06  1.452407e+06  \n",
       "mean       2.923434e+04  1.683267e+01  \n",
       "std        1.975315e+04  2.028013e+01  \n",
       "min        1.000000e+00  0.000000e+00  \n",
       "25%        1.043700e+04  6.000000e+00  \n",
       "50%        2.658800e+04  1.000000e+01  \n",
       "75%        4.755100e+04  1.900000e+01  \n",
       "max        6.536100e+04  2.070000e+02  "
      ]
     },
     "execution_count": 86,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train.describe()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 87,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\hdadmin\\Anaconda3\\lib\\site-packages\\sklearn\\grid_search.py:43: 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. This module will be removed in 0.20.\n",
      "  DeprecationWarning)\n"
     ]
    }
   ],
   "source": [
    "from sklearn import preprocessing\n",
    "from sklearn.cross_validation import train_test_split\n",
    "from sklearn.cross_validation import *\n",
    "from sklearn.grid_search import GridSearchCV"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 88,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "['app', 'os', 'channel', 'date_hour', 'ip_oscounts', 'app_clickip', 'app_os_clicks', 'app_device_click', 'clicks_by_ip', 'app_os_dev_count', 'app_rank']\n"
     ]
    }
   ],
   "source": [
    "features = list(train.columns) \n",
    "print(features)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 89,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "for f in train.columns:\n",
    "    if train[f].dtype=='object':\n",
    "        #print(f)\n",
    "        lbl = preprocessing.LabelEncoder()\n",
    "        lbl.fit(list(train[f].values) + list(test[f].values))\n",
    "        train[f] = lbl.transform(list(train[f].values))\n",
    "        test[f] = lbl.transform(list(test[f].values))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 90,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "xgb_model = xgb.XGBClassifier()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 91,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "## Set the params(this params from Pranav kernel) for xgboost model\n",
    "parameters = {'learning_rate': [0.2,0.3],\n",
    "          'booster': [\"gbtree\"],\n",
    "          'max_depth': [7,8],\n",
    "          'subsample': [1.0],\n",
    "          'colsample_bytree': [0.8],    \n",
    "          'min_child_weight':[0,1,20],\n",
    "          'reg_alpha':[5.0],\n",
    "          'gamma':[1.0],\n",
    "          'objective': ['binary:logistic'], \n",
    "          'scale_pos_weight':[1],\n",
    "          'nthread':[5],\n",
    "          'random_state': [99], \n",
    "          'silent': [1]}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 92,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "datetime.datetime(2018, 5, 7, 20, 50, 4, 173989)"
      ]
     },
     "execution_count": 92,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "import datetime\n",
    "datetime.datetime.now()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 93,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "clf = GridSearchCV(xgb_model, parameters, n_jobs=5, \n",
    "                   cv=StratifiedKFold(y, n_folds=5, shuffle=True), \n",
    "                   scoring='roc_auc',\n",
    "                   verbose=2, refit=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 94,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Fitting 5 folds for each of 12 candidates, totalling 60 fits\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[Parallel(n_jobs=5)]: Done  31 tasks      | elapsed: 44.3min\n",
      "[Parallel(n_jobs=5)]: Done  60 out of  60 | elapsed: 79.1min finished\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "GridSearchCV(cv=sklearn.cross_validation.StratifiedKFold(labels=[0 0 ..., 0 0], n_folds=5, shuffle=True, random_state=None),\n",
       "       error_score='raise',\n",
       "       estimator=XGBClassifier(base_score=0.5, booster='gbtree', colsample_bylevel=1,\n",
       "       colsample_bytree=1, gamma=0, learning_rate=0.1, max_delta_step=0,\n",
       "       max_depth=3, min_child_weight=1, missing=None, n_estimators=100,\n",
       "       n_jobs=1, nthread=None, objective='binary:logistic', random_state=0,\n",
       "       reg_alpha=0, reg_lambda=1, scale_pos_weight=1, seed=None,\n",
       "       silent=True, subsample=1),\n",
       "       fit_params={}, iid=True, n_jobs=5,\n",
       "       param_grid={'learning_rate': [0.2, 0.3], 'booster': ['gbtree'], 'max_depth': [7, 8], 'subsample': [1.0], 'colsample_bytree': [0.8], 'min_child_weight': [0, 1, 20], 'reg_alpha': [5.0], 'gamma': [1.0], 'objective': ['binary:logistic'], 'scale_pos_weight': [1], 'nthread': [5], 'random_state': [99], 'silent': [1]},\n",
       "       pre_dispatch='2*n_jobs', refit=True, scoring='roc_auc', verbose=2)"
      ]
     },
     "execution_count": 94,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "clf.fit(train[features], y)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 95,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "datetime.datetime(2018, 5, 7, 22, 11, 27, 55138)"
      ]
     },
     "execution_count": 95,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "datetime.datetime.now()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 96,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Raw AUC score: 0.9761115608713805\n",
      "booster: 'gbtree'\n",
      "colsample_bytree: 0.8\n",
      "gamma: 1.0\n",
      "learning_rate: 0.3\n",
      "max_depth: 8\n",
      "min_child_weight: 0\n",
      "nthread: 5\n",
      "objective: 'binary:logistic'\n",
      "random_state: 99\n",
      "reg_alpha: 5.0\n",
      "scale_pos_weight: 1\n",
      "silent: 1\n",
      "subsample: 1.0\n"
     ]
    }
   ],
   "source": [
    "#trust your CV!\n",
    "best_parameters, score, _ = max(clf.grid_scores_, key=lambda x: x[1])\n",
    "print('Raw AUC score:', score)\n",
    "for param_name in sorted(best_parameters.keys()):\n",
    "    print(\"%s: %r\" % (param_name, best_parameters[param_name]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 97,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "numpy.ndarray"
      ]
     },
     "execution_count": 97,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "type(clf.best_estimator_.feature_importances_)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 98,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.text.Text at 0x1af031a668>"
      ]
     },
     "execution_count": 98,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAA+sAAAGACAYAAADVrPP5AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAAIABJREFUeJzt3XuYHGWZ8P9vkglhMwQMMsKyq0IQb3BF5CAGNwi4goiA\n8XVXEdEFiRjPyoriCV1eWVdXRFRQQCLqri8qCoKHIKJICEQXXBWQ3Mri6YcHgkbAAIGE+f3x1Eg7\nzEw601Pp6pnv57pyZbqOd91dXV13PU9VTxscHESSJEmSJDXH9G4HIEmSJEmS/pLFuiRJkiRJDWOx\nLkmSJElSw1isS5IkSZLUMBbrkiRJkiQ1jMW6JEmSJEkN09ftACRJmmwiYhC4EVjfMvi6zFw0zuU9\nBTguMxdPRHyjrGMQGMjMO+paxyjrXQRslplnbcr1SpLUdBbrkiTV48AJLHz/DvjbCVpW0yygXNiQ\nJEktLNYlSdqEImJX4AzgkcAM4MOZuSQipgOnA/OBOcA0YBHwS+AUYKuI+CTwKeCjmfnEankHDL2O\niHcD+wJ/DfwoM4+OiLcDz6fc+vZz4FWZ+esx4tsB+Fb1b19gJvAm4BXALsB1wIuAxwDfAa4Edq/i\nfU1mLouImcAHgX+g9C74LvDGzLw7In5evX4S8DbgCOCgiLgXuBA4G9gW2A74BfCCzLy9mu/8apmP\nAT6XmW+uYn4Z8C/Vuu4A/jkzfxURhwPvADYD7gHelJnXjvkGSZLUEN6zLklSPb4dET9o+feoiOij\nFKQnZeZewP7AmyJiPvBUYHtg38x8AqUoPykzfwWcDCzLzGPbWO9jgT2rQv2lwG7APpn5ZOBrwCfa\nWMaOwCWZ+XfAFZSLCy+itPDvR7mgAKVovqxa9knA56pC/R3Vtuxe/ZsO/EfL8m/MzF0z8yLgEuD0\nzDwTOBK4NjP3BeZRCuyXtMy3RWbuBzwNeG1E7BgRuwPvAw7JzCdVy3t7ROwM/BtwaGbuARwPfCki\n+tvYfkmSus6WdUmS6vGwbvAR8QRgJ2BJRAwN/itgj8z8WES8A3hFROwEHADcPY71rsjMddXfhwH7\nANdV65sBzG5jGQ8Al1Z//y9wTWbeVW3Dr4GtgV8DqzPzswCZ+fWIWE9pMX828PbMfKCa5yPAxS3L\nXzbSSjPzjIjYLyJOAHYGnkhphR/y5Wq62yLi9iqO/SkXDH5VjftQtc5XUXoYXNGS6weBxwE/bCMH\nkiR1lcW6JEmbzgzgj1VLNAARsS1wZ0Q8h9KCfRqlKF0JHD3CMgYpXc6HbDZs/J+Gre99mfmxal2z\ngLltxHl/Zg62vH5glOnWDXs9ndIVfXjPvemU7vQjxfhnEfE+ysWFJcC3q3lat/Xelr+H8rCu+nto\nGX9F6V0wA7giM1/YMu7RlIsMkiQ1nt3gJUnadBK4LyKOhj8XjzcCewEHAZdWhfV/AwspBSeUgnSo\n2F0FPKbqVj+tmm40lwGLImLL6vUpwGcmcHsGIuKQalsOpxT1N1TrXRwRM6t78V8NXD7KMlq37VnA\nhzLzM8DtlJzMGGW+Id8GnhkRf129fgXwfso99wdHxC5VfIcCPwI23+itlCSpCyzWJUnaRDLzfuC5\nlAL6R8A3gHdm5nLg48D+1fBrKd3Pd6yK3WuBXSLiosz8MeUhbNcBK4DfjLHKTwBfAVZExE2ULurH\nTOAm3Qe8JCJ+CLwdWJiZ64H3AL8FfgDcTCnGXz/KMr4OvC4i3kq5mPCBiLge+BJwNaXb+qgy8wbg\nRGBpFcchwOLMvIlyn/oF1fD/CxyRmWs62WBJkjaVaYODgxueSpIkqUX11PgbM3OLbsciSdJkZMu6\nJEmSJEkNY8u6JEmSJEkNY8u6JEmSJEkNY7EuSZIkSVLDWKxLkiRJktQwfd0OoFOrVt3tTfdtmjt3\nNqtX39PtMHqW+euM+Rs/c9cZ89cZ89cZ89cZ89cZ89cZ8zd+5q59AwNzpo02zpb1KaSvb0a3Q+hp\n5q8z5m/8zF1nzF9nzF9nzF9nzF9nzF9nzN/4mbuJYbEuSZIkSVLDWKxLkiRJktQwFuuSJEmSJDWM\nxbokSZIkSQ1jsS5JkiRJUsNYrEuSJEmS1DC1/c56REwHzgJ2B9YCizLzlmHTzAYuB47LzJXVsLcC\nRwCbAWdl5nl1xShJkiRJUhPV2bK+ENg8M/cFTgJOax0ZEXsDVwE7tQw7AHga8PfA/sCja4xPkiRJ\nkqRGqq1lHVgALAXIzBVVcd5qFvA84DMtw54F3ABcBGwJnLihlcydO5u+vhkTEvBUMDAwp9sh9DTz\n1xnzN37mrjPmrzPmrzPmrzPmrzPmrzPmb/zMXefqLNa3BO5seb0+Ivoycx1AZi4HiIjWebYBHgsc\nBuwIXBIRu2Tm4GgrWb36nomOe9IaGJjDqlV3dzuMnmX+OmP+xs/cdcb8dcb8dcb8dcb8dcb8dcb8\njZ+5a99YFzXqLNbvAlrXPH2oUB/D74GVmXk/kBFxHzAA3F5TjJIkSZIkNU6d96wvBw4FiIj5lO7t\nG3I1cEhETIuI7YF+SgEvSZIkSdKUUWexfhFwX0RcA5wOvDEijoqI40ebITO/AvwP8D3gUuDVmbm+\nxhglSZIkSWqc2rrBZ+aDwOJhg1eOMN0Bw16/ua6YJEmb3sXLbu3q+vv7Z7FmzdqurX/hfvO6tm5J\nktS76mxZlyRJkiRJ42CxLkmSJElSw9T5NHhJkiRpyvI2IG8Dkjphy7okSZIkSQ1jsS5JkiRJUsNY\nrEuSJEmS1DAW65IkSZIkNYzFuiRJkiRJDWOxLkmSJElSw1isS5IkSZLUMBbrkiRJkiQ1jMW6JEmS\nJEkNY7EuSZIkSVLDWKxLkiRJktQwFuuSJEmSJDWMxbokSZIkSQ1jsS5JkiRJUsNYrEuSJEmS1DAW\n65IkSZIkNYzFuiRJkiRJDWOxLkmSJElSw1isS5IkSZLUMBbrkiRJkiQ1jMW6JEmSJEkNY7EuSZIk\nSVLDWKxLkiRJktQwFuuSJEmSJDWMxbokSZIkSQ1jsS5JkiRJUsNYrEuSJEmS1DAW65IkSZIkNUxf\nXQuOiOnAWcDuwFpgUWbeMmya2cDlwHGZubJl+KOA64GDWodLkiRJkjQV1NmyvhDYPDP3BU4CTmsd\nGRF7A1cBOw0bPhM4G7i3xtgkSZIkSWqsOov1BcBSgMxcAew9bPws4HnA8JbzDwAfB35dY2ySJEmS\nJDVWbd3ggS2BO1ter4+IvsxcB5CZywEi4s8TRMQxwKrMvCwi3trOSubOnU1f34wJC3qyGxiY0+0Q\nepr564z5G79ezl1//6xuh9DVGHr5vRsyGbahm8xfZ3o5fx7/eve9GzIZtqFbzF3n6izW7wJa36Hp\nQ4X6GF4GDEbEM4EnA5+OiCMy87ejzbB69T2dRzpFDAzMYdWqu7sdRs8yf50xf+PX67lbs2ZtV9ff\n3z+rqzH08nsHvb//dZv560yv58/jX+++d9D7+183mbv2jXVRo85ifTlwOPD5iJgP3LChGTLz6UN/\nR8SVwOKxCnVJkiRJkiajOov1i4CDIuIaYBpwbEQcBWyRmefUuF5JkiRJknpabcV6Zj4ILB42+GE/\nw5aZB4wy/4jDJUmSJEma7Op8GrwkSZIkSRoHi3VJkiRJkhrGYl2SJEmSpIaxWJckSZIkqWEs1iVJ\nkiRJahiLdUmSJEmSGsZiXZIkSZKkhrFYlyRJkiSpYSzWJUmSJElqGIt1SZIkSZIaxmJdkiRJkqSG\nsViXJEmSJKlhLNYlSZIkSWoYi3VJkiRJkhrGYl2SJEmSpIaxWJckSZIkqWEs1iVJkiRJahiLdUmS\nJEmSGsZiXZIkSZKkhrFYlyRJkiSpYSzWJUmSJElqGIt1SZIkSZIaxmJdkiRJkqSGsViXJEmSJKlh\nLNYlSZIkSWoYi3VJkiRJkhrGYl2SJEmSpIaxWJckSZIkqWEs1iVJkiRJahiLdUmSJEmSGsZiXZIk\nSZKkhumra8ERMR04C9gdWAssysxbhk0zG7gcOC4zV0bETGAJsAMwC3hPZl5SV4ySJEmSJDVRnS3r\nC4HNM3Nf4CTgtNaREbE3cBWwU8vgo4HfZ+Z+wCHAR2uMT5IkSZKkRqqzWF8ALAXIzBXA3sPGzwKe\nB6xsGfYF4J3V39OAdTXGJ0mSJElSI9XWDR7YEriz5fX6iOjLzHUAmbkcICL+PEFm/qkaNge4EHjH\nhlYyd+5s+vpmTGDYk9vAwJxuh9DTzF9nzN/49XLu+vtndTuErsbQy+/dkMmwDd1k/jrTy/nz+Ne7\n792QybAN3WLuOldnsX4X0PoOTR8q1McSEY8GLgLOyszPbmj61avvGX+EU8zAwBxWrbq722H0LPPX\nGfM3fr2euzVr1nZ1/f39s7oaQy+/d9D7+1+3mb/O9Hr+PP717nsHvb//dZO5a99YFzXq7Aa/HDgU\nICLmAzdsaIaI2Bb4BvCWzFxSY2ySJEmSJDVWnS3rFwEHRcQ1lPvPj42Io4AtMvOcUeZ5GzAXeGdE\nDN27/uzMvLfGOCVJkiRJapTaivXMfBBYPGzwyhGmO6Dl79cDr68rJkmSJEmSekGd3eAlSZIkSdI4\nWKxLkiRJktQwFuuSJEmSJDWMxbokSZIkSQ1jsS5JkiRJUsNYrEuSJEmS1DAW65IkSZIkNYzFuiRJ\nkiRJDWOxLkmSJElSw1isS5IkSZLUMBbrkiRJkiQ1jMW6JEmSJEkNY7EuSZIkSVLDWKxLkiRJktQw\nFuuSJEmSJDVMX7cDkLRpXLzs1q6uv79/FmvWrO3a+hfuN69r65YkSZI2li3rkiRJkiQ1jMW6JEmS\nJEkNY7EuSZIkSVLDWKxLkiRJktQwFuuSJEmSJDWMxbokSZIkSQ1jsS5JkiRJUsNYrEuSJEmS1DB9\n3Q5AknrBxctu7dq6+/tnsWbN2q6tf+F+87q2bkmSpKnKlnVJkiRJkhrGYl2SJEmSpIaxWJckSZIk\nqWEs1iVJkiRJahiLdUmSJEmSGsZiXZIkSZKkhrFYlyRJkiSpYWr7nfWImA6cBewOrAUWZeYtw6aZ\nDVwOHJeZK9uZR5IkSZKkya7OlvWFwOaZuS9wEnBa68iI2Bu4Ctip3XkkSZIkSZoK6izWFwBLATJz\nBbD3sPGzgOcBKzdiHkmSJEmSJr3ausEDWwJ3trxeHxF9mbkOIDOXA0RE2/OMZO7c2fT1zZi4qCe5\ngYE53Q6hp/Vy/vr7Z3U7hK7G0Ol71+38mbvO9HL+mmAybEM3mb/O9HL+PP717ns3ZDJsQ7eYu87V\nWazfBbS+Q9PHKrrHO8/q1feMM7ypZ2BgDqtW3d3tMHpWr+dvzZq1XV1/f/+srsbQ6XvXzdjNXWd6\nPX/d1uvHvm4zf53p9fx5/Ovd9w56f//rJnPXvrEuatTZDX45cChARMwHbqhpHkmSJEmSJpU6W9Yv\nAg6KiGuAacCxEXEUsEVmntPuPDXGJ0mSJElSI9VWrGfmg8DiYYNXjjDdARuYR5IkSZKkKaXObvCS\nJEmSJGkcLNYlSZIkSWoYi3VJkiRJkhrGYl2SJEmSpIZp6wFzEbEZcCIQwGuANwD/npn31xibJEmS\nJElTUrst62cC/cCewDrgccB5dQUlSZIkSdJU1m6xvldmvg14IDPvAf4Z2KO+sCRJkiRJmrraLdYH\nq67wg9XrbVr+liRJkiRJE6jdYv1DwDeB7SLiQ8D1wOm1RSVJkiRJ0hTW1gPmMvMzEXE9cCAwA3hO\nZt5Qa2SSJEmSJE1RbbWsR8RuwMmZeSZwOXBmREStkUmSJEmSNEW12w3+XOB8gMy8Gfi/+DR4SZIk\nSZJq0W6x3p+ZS4deZObllJ9ykyRJkiRJE6yte9aB2yNiMfCf1esjgd/VE5IkSZIkSVNbuy3rxwKH\nAb8Bfgk8B1hUV1CSJEmSJE1l7T4N/peUYl2SJEmSJNWsrWI9Ip4FvAfYGpg2NDwz59UUlyRJkiRJ\nU1a796x/BDgBuBEYrC8cSZIkSZLUbrF+R2Z+pdZIJEmSJEkS0H6xviwiPggsBe4bGpiZV9USlSRJ\nkiRJU1i7xfo+1f97tAwbBJ4xseFIkiRJkqR2nwZ/YN2BSJIkSZKkot2nwS8ATgS2oDwNfgbw2Mzc\nob7QJEmSJEmamqa3Od0ngIspxf2ZwE+Bi+oKSpIkSZKkqazde9bvzcxPRsQOwGrg5cD1tUUlSZIk\naUq7eNmtXV1/f/8s1qxZ27X1L9xvXtfWrWZot2X9vojYGkhgfmYOAv31hSVJkiRJ0tTVbrH+QeBz\nwKXASyPiJuC62qKSJEmSJGkKa7cb/DeBCzNzMCL2Ah4P/LG+sCRJkiRJmrrGLNYj4tGUp79/DXh2\nREyrRt0JfB3Ypd7wJEmSJEmaejbUsv6vwIHA9sBVLcMfAL5aV1CSJEmSJE1lYxbrmfkygIh4S2a+\nb9OEJEmSJEnS1NbuPevHABtVrEfEdOAsYHdgLbAoM29pGX84cDKwDliSmedGxEzgU8AOwHrg5Zm5\ncmPWK0mSJElSr2u3WP9xRJwMfBe4d2hgZl41+iwsBDbPzH0jYj5wGvBcgKooPx14CrAGWB4RlwDz\ngb7MfFpEHAScCjx/I7dJk5S/telvbUqSJElTRbvF+taUe9cPbBk2CDxjjHkWAEsBMnNFROzdMm5X\n4JbMXA0QEVcDTwduBPqqVvktKffGS5IkSZI0pbRVrGfmgQARMQeYkZnt/GzblpSnxg9ZHxF9mblu\nhHF3A1sBf6J0gV8JbAMctqGVzJ07m76+Ge1shoCBgTndDmHc+vtndTuErsbQ6Xtn/no7f+auM72c\nvyaYDNvQTeavM72cP49/fn90opf3fej9+JugrWI9IuYBFwA7AdMi4hfACzLzp2PMdhfQ+g5Nrwr1\nkcbNofxu+xuByzLzrdXPxn0rInbLzPtGW8nq1fe0swmifGBWrbq722GMWze7oEP3u8F3+t6Zv97N\nn7nrTK/nr9t6/buj28xfZ3o9fx7//P7oRC/v+73+2d2UxrqoMb3NZZwNvD8zH5mZWwPvBc7dwDzL\ngUMBqnvWb2gZdzOwc0RsHRGbUbrAXwus5qEW9z8AMwGbzSVJkiRJU0q7xfo2mXnh0IvM/DzlPvax\nXATcFxHXUB4m98aIOCoijs/MB4ATgMsoRfqSzLytmm7PiFgGfAt4W2au2bhNkiRJkiSpt7X7gLm1\nEbFnZn4fICL2Asbsf56ZDwKLhw1e2TL+UuDSYfP8CXhBmzFJkiRJkjQptVusvwH4YkT8AZhGaVV/\nYW1RSZIkSZI0hbX7NPgVEfF44PGUYv0nmXl/rZFJkiRJkjRFtXXPekQ8BrgQWAFcBSyJiIE6A5Mk\nSZIkaapq9wFz/wVcDmwP7AhcD3yqrqAkSZIkSZrK2r1nfcvM/GjL69Mj4pga4pEkSZIkacprt2X9\n+og4euhFRDwH+J96QpIkSZIkaWprt2X9MOCYiDgbGARmA0TES4HBzJxRU3ySJEmSJE057T4N/lF1\nByJJkiRJkoq2ivXqye9HAnNbh2fmKXUEJUmSJEnSVNbuPetfA/ag/MZ66z9JkiRJkjTB2r1nncx8\nWZ2BSJIkSZKkot1i/eKIWAR8C1g3NDAzf1lLVJIkSZIkTWHtFutbAScBd7QMGwTmTXhEkiRJkiRN\nce0W688HHpWZ99YZjCRJ0kS6eNmtXV1/f/8s1qxZ27X1L9zPdhVJ6lXtFuu3Up4Eb7EuSZI0RXix\nw4sdkrqn3WJ9EPhxRNwI3D80MDOfUUtUkiRJkiRNYe0W66fWGoUkSZIkSfqztor1zPxO3YFIkiRJ\nkqRizGI9Ih6kdIEfbhowmJkzaolKkiRJkqQpbMxiPTOnb6pAJEmSJElSYTEuSZIkSVLDWKxLkiRJ\nktQwFuuSJEmSJDWMxbokSZIkSQ1jsS5JkiRJUsNYrEuSJEmS1DAW65IkSZIkNYzFuiRJkiRJDWOx\nLkmSJElSw1isS5IkSZLUMBbrkiRJkiQ1jMW6JEmSJEkN01fXgiNiOnAWsDuwFliUmbe0jD8cOBlY\nByzJzHOr4W8FjgA2A87KzPPqilGSJEmSpCaqrVgHFgKbZ+a+ETEfOA14LkBEzAROB54CrAGWR8Ql\nwK7A04C/B2YDb6oxPkmSJEmSGqnObvALgKUAmbkC2Ltl3K7ALZm5OjPvB64Gng48C7gBuAi4FPhK\njfFJkiRJktRIdbasbwnc2fJ6fUT0Zea6EcbdDWwFbAM8FjgM2BG4JCJ2yczB0VYyd+5s+vpmTHjw\nk9XAwJxuhzBu/f2zuh1CV2Po9L0zf72dP3PXmV7OXxP08ja4//n57YT564z560wvH3uh9+NvgjqL\n9buA1ndoelWojzRuDvBH4PfAyqq1PSPiPmAAuH20laxefc+EBj2ZDQzMYdWqu7sdxritWbO2q+vv\n75/V1Rg6fe/MX+/mz9x1ptfz121+d3Sm1/c/82f+OmH+OtPLx95e/+7YlMa6qFFnN/jlwKEA1T3r\nN7SMuxnYOSK2jojNKF3gr6V0hz8kIqZFxPZAP6WAlyRJkiRpyqizZf0i4KCIuAaYBhwbEUcBW2Tm\nORFxAnAZ5YLBksy8DbgtIp4OfK8a/urMXF9jjJIkNdrFy27t6vq73bK0cL95XVu3JEndVFuxnpkP\nAouHDV7ZMv5SykPkhs/35rpikiRJkiSpF9TZDV6SJEmSJI2DxbokSZIkSQ1jsS5JkiRJUsNYrEuS\nJEmS1DAW65IkSZIkNYzFuiRJkiRJDWOxLkmSJElSw9T2O+t6uIuX3drV9ff3z2LNmrVdW//C/eZ1\nbd2SJEmS1EtsWZckSZIkqWEs1iVJkiRJahiLdUmSJEmSGsZ71iVJkiRpEvFZWZPjWVm2rEuSJEmS\n1DAW65IkSZIkNYzFuiRJkiRJDWOxLkmSJElSw1isS5IkSZLUMBbrkiRJkiQ1jMW6JEmSJEkNY7Eu\nSZIkSVLDWKxLkiRJktQwFuuSJEmSJDWMxbokSZIkSQ1jsS5JkiRJUsNYrEuSJEmS1DAW65IkSZIk\nNYzFuiRJkiRJDWOxLkmSJElSw1isS5IkSZLUMBbrkiRJkiQ1jMW6JEmSJEkN01fXgiNiOnAWsDuw\nFliUmbe0jD8cOBlYByzJzHNbxj0KuB44KDNX1hWjJEmSJElNVGfL+kJg88zcFzgJOG1oRETMBE4H\nDgb2B46PiG1bxp0N3FtjbJIkSZIkNVadxfoCYClAZq4A9m4ZtytwS2auzsz7gauBp1fjPgB8HPh1\njbFJkiRJktRYtXWDB7YE7mx5vT4i+jJz3Qjj7ga2iohjgFWZeVlEvLWdlcydO5u+vhkTFXOt+vtn\ndTuErsYwMDCno/nNn/nrRK/nz9x1xvx1xvx1xvx1xvx1xvx1plfzZ+462/eaos5i/S6gNUvTq0J9\npHFzgD8CrwMGI+KZwJOBT0fEEZn529FWsnr1PRMbdY3WrFnb1fX398/qagyrVt3d0fzmz/x1opfz\nZ+46Y/46Y/46Y/46Y/46Y/4608v5M3ed7Xub0lgXFuos1pcDhwOfj4j5wA0t424Gdo6IrYE/UbrA\nfyAzLxyaICKuBBaPVahLkiRJkjQZ1VmsXwQcFBHXANOAYyPiKGCLzDwnIk4ALqPcN78kM2+rMRZJ\nkiRJknpGbcV6Zj4ILB42eGXL+EuBS8eY/4B6IpMkSZIkqdnqfBq8JEmSJEkaB4t1SZIkSZIaxmJd\nkiRJkqSGsViXJEmSJKlhLNYlSZIkSWoYi3VJkiRJkhrGYl2SJEmSpIaxWJckSZIkqWEs1iVJkiRJ\nahiLdUmSJEmSGsZiXZIkSZKkhrFYlyRJkiSpYSzWJUmSJElqGIt1SZIkSZIaxmJdkiRJkqSGsViX\nJEmSJKlhLNYlSZIkSWoYi3VJkiRJkhrGYl2SJEmSpIaxWJckSZIkqWEs1iVJkiRJahiLdUmSJEmS\nGsZiXZIkSZKkhrFYlyRJkiSpYSzWJUmSJElqGIt1SZIkSZIaxmJdkiRJkqSGsViXJEmSJKlhLNYl\nSZIkSWoYi3VJkiRJkhrGYl2SJEmSpIbpq2vBETEdOAvYHVgLLMrMW1rGHw6cDKwDlmTmuRExE1gC\n7ADMAt6TmZfUFaMkSZIkSU1UZ8v6QmDzzNwXOAk4bWhEVZSfDhwM7A8cHxHbAkcDv8/M/YBDgI/W\nGJ8kSZIkSY1UZ7G+AFgKkJkrgL1bxu0K3JKZqzPzfuBq4OnAF4B3VtNMo7S6S5IkSZI0pdTWDR7Y\nEriz5fX6iOjLzHUjjLsb2Coz/wQQEXOAC4F3bGglc+fOpq9vxsRFXaP+/lndDqGrMQwMzOlofvNn\n/jrR6/kzd50xf50xf50xf50xf50xf53p1fyZu872vaaos1i/C2jN0vSqUB9p3BzgjwAR8WjgIuCs\nzPzshlayevU9ExPtJrBmzdqurr+/f1ZXY1i16u6O5jd/5q8TvZw/c9cZ89cZ89cZ89cZ89cZ89eZ\nXs6fuets39uUxrqwUGexvhw4HPh8RMwHbmgZdzOwc0RsDfyJ0gX+A9V9698AXpOZV9QYmyRJkiRJ\njVVnsX4RcFBEXEO5//zYiDgK2CIzz4mIE4DLKPfNL8nM2yLiDGAu8M6IGLp3/dmZeW+NcUqSJEmS\n1Ci1FeuZ+SCweNjglS3jLwUuHTbP64HX1xWTJEmSJEm9oM6nwUuSJEmSpHGwWJckSZIkqWEs1iVJ\nkiRJahiLdUmSJEmSGsZiXZIkSZKkhrFYlyRJkiSpYSzWJUmSJElqGIt1SZIkSZIaxmJdkiRJkqSG\nsViXJEmSJKlhLNYlSZIkSWoYi3VJkiRJkhrGYl2SJEmSpIaxWJckSZIkqWEs1iVJkiRJahiLdUmS\nJEmSGsZiXZIkSZKkhrFYlyRJkiSpYSzWJUmSJElqGIt1SZIkSZIaxmJdkiRJkqSGsViXJEmSJKlh\nLNYlSZIkSWoYi3VJkiRJkhrGYl2SJEmSpIaxWJckSZIkqWEs1iVJkiRJahiLdUmSJEmSGsZiXZIk\nSZKkhrFYlyRJkiSpYSzWJUmSJElqGIt1SZIkSZIapq+uBUfEdOAsYHdgLbAoM29pGX84cDKwDliS\nmeduaB5E3oQWAAARfklEQVRJkiRJkqaCOlvWFwKbZ+a+wEnAaUMjImImcDpwMLA/cHxEbDvWPJIk\nSZIkTRV1FusLgKUAmbkC2Ltl3K7ALZm5OjPvB64Gnr6BeSRJkiRJmhKmDQ4O1rLgiPgE8MXM/Hr1\n+pfAvMxcFxELgNdm5gurcacAvwTmjzZPLUFKkiRJktRAdbas3wXMaV1XS9E9fNwc4I8bmEeSJEmS\npCmhzmJ9OXAoQETMB25oGXczsHNEbB0Rm1G6wF+7gXkkSZIkSZoS6uwGP/Rk9ycB04BjgT2BLTLz\nnJanwU+nPA3+zJHmycyVtQQoSZIkSVJD1VasS5IkSZKk8amzG7wkSZIkSRoHi3VJkiRJkhrGYl3S\nRomIYyLi3zfh+t4dEYs31fomSkRsHhE/H2P88RExcyOXuUNErOg0tqaIiEMi4vhux7Exqvd1Ubfj\n6IbW/S8iLqgeEDvSdOdHxCHDhp0UEftsijh7XURcGRG7VMfaI0aZ5oCIuGBTx9apiFgRETts5Dxf\nqimcP++rYx2LNsVxd+h7NSK2i4izxpju5xGx+TjX0fF21PleNE1ELI6Id3c7jrGM5zyil/Tq+d9E\n6+t2AJI0Rb0N+DTwQLcD6ZbMXNrtGMZhO2AR8IluB9JNmXnkRk6/yS7wTRaZeX63Y2iCzPw/m2Ad\njTgWZeZvgVd1O47RbIr3Qhtlyp9HTAUW65NIRGxJOYF8BLA9cCbwQmAlsAvlCfsvrP5+O/Ag5cTz\nnMw8sxsxN1V1pfKTwDxgBvBB4JHAP1Py9t+Z+bruRbjpRMRfUXLxWGAz4EJgfkR8AxgAPlb9wsM/\nAq8GZgKDwPOAJwJvAe6n5PKCzDw1Is4H1gI7AH8NHJOZ34+IfwJOANYDV2fmSZtsQydARGwB/Bcw\nF7ilGrY/8C5KT6YtgKOA/SifvQuAhRHx3mrYDOCDmfmFMVYzEBEXU/L2o8x8edVStYRyTB8EXpeZ\nP4yI32bmdlUcFwAfp+T8ZVU878rMKyYuAxsnIo4BDgF2An4D/C3w9cx8+xjzvBh4A2X/+SlwPLAj\nZR9dR9muo4D/D/gIsA9lv31XZn45Ik4DFlSL+2xmnlHtjxdk5tKqRfjIzDwmIn5K+UnRAH4HPJ9y\n7HxCRJwMXAGcRjlRugf4x8y8eyJyszHqOvZHxDuAhZT96mPAZS3jfl4t79HVujej5ODIlmmeCnwY\n+CfgFMr+vl21zDnANsApmfnFjhLQhhpz9C+UbV4HXJWZb4mIv6fN/aLK0Yco++1twItbxr0b+C1w\nNsP2ZeDOaprZwBeB/wS+AXyuWtbmwOLM/EEDcnQq5XP+K8p7TkRsBZxH+V4FeB3lc/y8zDy2mub7\n1Xw/ysztRsnV4yj72DTg98DLMvPOUeLYmdH31WOAXTLzpNH2+4iYAZwP3NTpxadRvlepjuUXZOb8\niDiM8l5PA74PLG6ZfzFwMPAiyi8rHVjF+8XMfN8Yqx6IiEuAbYGvAKcCPwH2ycw/RMQrgTmZ+f5R\n4v5t9V5cybD9orrQ0O72N/XzuAA4A1hdLWOoJ9FrKd8rg5Tj2McoP0O9e2auiYg3Aesz8/RRlnsc\n8ErKd/wlmfmuUb7LXsxD++HmwMrM3KHK9w8o51RbUo6pz6TlPGK0nGyMOt6Xap++lPL5/BrwXR5+\nTnQ/8P8ox4idgO9l5itblvE44LPAosz80URsay+xG/zk8jjKQf5gykH8hGr4NZl5AOVL/G3VsL8B\njgDmA2+MiEdt4lib7hXAqsx8GuWA+B7gROA1mbkvcHNETJWLXYuBn1fbfSRwL+VL71mUgvwN1XSP\nB56TmQuAH1fjoZyMPJ+yr725Zbm/yMxnUU5Cj4+IrYF/Bf6hWsbfRMRBtW7ZxFsM3JiZT6ecYAP8\nHXB09Rn8EvBPmXke5ST8yIh4NrBjtc0HAm+PiEeMsY4tKT+FuS/wD9Vn9wPAGdV6X085CR7L6sxc\n0M1CfZgdgGOApwDPiIg9R5ooIh5J2UeeUeXrj5TP6kHA9yif1XcBW1FOXrbJzH0oed27OvndkbIv\nLgCOiojdxohrHvDOat8fqOI7FfhxZp5SrePzwP6Uk7e549z+Tk34sT8i9gCeDTyVUiQ+nnKiNtwH\ngPdWOToD2KMa/jTKRc7DM/OXw+bpp7xnBwMf3ETH0jpytBvwAsq2Pg3YudrHNma/OJtSYD4V+Cqw\n6wjTPGxfroZvQTkJ/lhm/hflffo95X17NSXPG6OOHO0NPJ3y2Xkp5SIN1XKuyMwDKUXKx6rt3zci\n+iPiKcCtmXl7y+JGytW5wKur+L7GX37HDDfavtoa72j7fR/lQuy1E9RLZKTv1dY4+oCPUr5T96Zc\n/P3bavRrKRd3/ykz11IKvKGLwH/cwHq3AF5C2V+fDexWbdfQhYujgU+1uQ0j7Rftaurn8WPAizLz\nmcDPquU+gVKgLqDkeCHlu+GLlHMbKPn/9ChxPQo4qZp3T2BWRDyWkb/LxvK9Kq7Lqxj/fB6xgfk2\nRl11xHbAwdVFoIedE1XTPB44jvK5OzQitquGB6VQf/FULNTBYn2y+R2lle4/gXdQWjgBvlX9fw1l\np4fywVubmfcCN1KuZOkhuwJXAVRXYH9MKSZeHRHfoRSgI524TkYBXAuQmT+lfKl8PzMHKV8Us6vp\nbgc+FRGfBJ7EQ/vfDZm5LjPX8JcnJP9T/f8rSivQ4ygF0deqq8hPoPf2y8dTikYy87uUixq3AR+u\nWm8P5KG8DNkN2Kva5qXV+B3GWMetmbk6Mx+k5Hw2f7m//oDS2jlc6/6aG7NRm8APM/MPmbmectU9\nRpluHqVVa6hV5CrKF/95lP1yKfAaSotI6367OjPfScnTsswczMwHKK0mTxi2jtY83ZGZv6r+HtpP\nW/0bpfXhCuAf6V5XxDqO/UE5OVyfmfdn5r9QWpVGmm4oz5dk5jeq4QdTWmdGysl3MvPBzPwdpQVr\noN0N7UAdOdoFWJGZD1THw2WU/XFj9ovtMvNmgMw8LzO/P8I0I+3LUIqPvwJmVa+/TukJ8mVKT4YH\nx1jvSOrI0eOB66r3+y7ghmr4bsDLquPeucDW1ef/QuD/UC5InjtsWSPlalfgrGo5L6MUEKMZbV8d\nPs1I+/3uwKMoxe5EGOl7tdU2lIuqt1fTvL/lotczgUdU+YJSrP87pQfAWBd6oRxr76zm/R7l/VkC\nvCQingj8rvpctmOk/aJdTf08bpuZP6n+Xl79/0TKOd8V1b9HAkO9NF4a5VkcmZm/H2WZ8ygX8e+t\nvntOouxLI32XtRp+jjn8nKkOddURP8vM+6u/RzsnuiUz7672zd/w0DY+m3Kes54pymJ9cvkXylXf\no4Ev8NAHfa/q/78Hbqr+fnJEzKi60P0dpQuOHnIz5SooETGHcmJxJKVb4f6UK/JP6154m9TNlFYR\nImIe5YvvL07aqy6N/0rJ0SJKUT60/410gj/S8J9RvoQOqq64foSqC1oP+TGlxXuohWYm5YTz2Mw8\nBvg1D+XlQcoxeCXw7Wqbn0FpAfjfMdYxUj5b99cnUy6iAMyMiC2iPAis9URgY0/i67ZrRMyuupk+\nlZLHkfyM0gV9qMVwf0oXzudSivB/oBz73sJf7rdbRcRl1bAF1bCZlM/wT4H7KLcVQGn5GDJSrofe\nNyitUOdXrYM3UVoIu6GOY/9KYM+ImB4RMyPich4qClu15vnFVXdRgHcDpwMjPSxrr2r6bSk9RW4f\nYZqJVleOnhoRfRExjdKC/BM2br/4ddU9m4h4S0Q8b4RpRtqXobQuPw84NSK2Bw4AflO1ir2Hcqze\nGHXk6MfAPtV+1M9DF8dWAqdXx70XULrxQ7nw9hLKceDyYcsaKVcJvLRazpspXbtHM9q+2mq0/f56\n4DmUovZJY6yjXSN9r7a6HXhE1eOMiPhwPPSAxucCq6M8AG0WpWXyRZTC55iq1XY0u1bfCX2UHN+U\nmb+gXCx4OxvuldVqpP2iXU39PN4WEUO9W55S/Z/VfAdW+9n5lFszflrFfSIPv7DU6n+BXar3ioi4\nkPL+jvRdNtp3EWz4+2gi1FVHtJ5zjHZONNq54oeAN1Iag2ZsxLZMGhbrk8ulPNTy+wZK69IsysH7\nO5QvmlOraWdSrsIvA96TmXd0Id4mOwd4ZERcDVxJKUSvB5ZFxLcoB9rvdi+8TepsYF61D32a0rV1\nuLsoV6GvpexT91KuZLctM1dVy/5ORHyXcjX1J2PP1Tgfp+Tqako31LWUk9BlEbGc0gV0KC/LKN02\nLwX+FBHLKPvYYG78fc9vAl4bEVdRuvEdVw3/EOWCx4XAL8a9VfW7n3Ji8F3gy5n5w5Emqo5T7wK+\nHeWpxttQtvc64JTqs7mYcqHnEsoJ7dWUFqcPZeZXgJ9FxLVUeala5z5B6cb3TcZumYPy2d8sIt5H\naZn6RERcQbnQMmI3yE1gwo/9VQ+NpZTP9dWUrrJrR5j0ROCtVcvmi6vphpbxCWDriDhq2DzbVTn7\nKvCqlhbCOtWRoxsoF9eWU/aFnwMXs3H7xSuAJVUMe1COCcM9bF9uieF3lM/EJ4EfAouq9+I/gPeO\nsd6R1LUffR34b8q9tUMXZk4FXtDSo+jGavqfVeO/XPUeajVSrl4JfLrKzb8DY3WTHXVfHRbviPt9\n1YI4tL6RLlxtjDG/V6ttfxXw1WrbplFyOOR1lOP+Y4A/UI5n36Y8t2D4bSet/kDpynwN5fg3dGH0\nXMoF34150N5I+0W7mvx5/HQ17WOr5f6Q0qJ+dURcR2lVv62a/jzKvvjt0RZYndu8j3Jucy2lZ+Iv\nGPm7bCmwQ/Wev4BybjWWZZTeiBPV03NT1BGjnRONKjMvp1z4e8tGbMukMW1wcLQLGZoMqi+lxZm5\nsmXYAdWwibzPRZI2WrQ8UKnbsUwmTT32R8uDvLoVQ0ssV9LAHDWJOZoaojzcdbfMPLnN6a9k2H4x\nATE8bJnua93n+9J9U+UBWZLUM6L83u/w1kiAt2bmtZs6nk0tym9LnzDCqDMy86JNHc9kNtX3tXZ0\nmqOIeAwjt+Z9JzPf1Wl8TdCU/SjKLT8j3YuembmhB3hNChv7XkTEv1G60B9WvW708bepn8em561u\nTTkGTEa2rEuSJEmS1DDesy5JkiRJUsNYrEuSJEmS1DAW65IkSZIkNYwPmJMkaZKonq7/Ex7+W/WH\nZ+avNmI5OwLvyMzjNjixJEmqhcW6JEmTy68z88kdLuOxwE4TEYwkSRofi3VJkia5iNgWOBt4NPAg\n5ed0vhkRfwOcBzwC+Gvg/1W/gf5hYF5EnAl8AXh3Zh5QLet84Mrq31LgDuA+4FnAfwAHADOA8zPz\n9Ij4W+C/gP5q3a/LzBW1b7QkST3Oe9YlSZpcto+IH7T8OxE4A1iSmXsBRwBnR8Qc4EWUAn0+8CTg\nVRGxDfA64LrMfPUG1hXA0Zn5TODlAJm5J7AP8NyI2A84DvhKZu4NvBlYMOFbLEnSJGTLuiRJk8vD\nusFHxB3ALhFxSjVoJrBTZn4gIg6MiDcBTwQ2o7SAt+v2zPx59fczgSdHxDOq11sAuwHfBL4UEXsA\nXwU+Op6NkiRpqrFYlyRp8psBPCMz/wAQEdsDv4uI04B5wGeBiykF97Rh8w4OGzaz5e97h63jzZn5\npWod2wBrMvPeiHgCcBjwQuAY4KAJ2i5JkiYtu8FLkjT5fQt4FUBVOP8ImE0pmv8jM79AuZ/9byhF\n9zoeuqB/B+X+9c0jYmtgvzHW8fKImBkRWwBXA0+NiPcDL8nMTwGvAfasYwMlSZpsLNYlSZr8XgvM\nj4gfAZ+jFM93A+8FPhMR1wMnAtcBOwI3A4+IiM9k5k2U7us3UR42t2yUdXwc+CnwP9VyPpmZVwIf\nAZ4fET8ALgJeWc8mSpI0uUwbHBzsdgySJEmSJKmFLeuSJEmSJDWMxbokSZIkSQ1jsS5JkiRJUsNY\nrEuSJEmS1DAW65IkSZIkNYzFuiRJkiRJDWOxLkmSJElSw1isS5IkSZLUMP8/qrhAdr+Z6+gAAAAA\nSUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x1b5f5f62b0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "y_pos = np.arange(len(features))\n",
    "performance = clf.best_estimator_.feature_importances_\n",
    "plt.figure(figsize=(17,6)) \n",
    "plt.bar(y_pos, performance, align='center', alpha=0.5)\n",
    "plt.xticks(y_pos, features)\n",
    "plt.ylabel('Importance')\n",
    "plt.xlabel('Features')\n",
    "plt.title('Feature Importance')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 99,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "test_probs = clf.predict_proba(test[features])[:,1]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 100,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "sub['is_attributed'] = np.round(test_probs).astype(np.int64)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 101,
   "metadata": {
    "scrolled": true
   },
   "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>click_id</th>\n",
       "      <th>is_attributed</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>3</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>4</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   click_id  is_attributed\n",
       "0         0              0\n",
       "1         1              0\n",
       "2         2              0\n",
       "3         3              0\n",
       "4         4              0"
      ]
     },
     "execution_count": 101,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "sub.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 102,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "sub.to_csv(path+'xgb_sub.csv',index=False)"
   ]
  },
  {
   "cell_type": "raw",
   "metadata": {
    "collapsed": true
   },
   "source": [
    "part 1 25 = 0 - 46225951\n",
    "part 2 50 = 46225951 - 92451901\n",
    "part 3 75 = 92451901 - 138677851\n",
    "part 4 100 = 138677851 - 184903890"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 103,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "part1 = 0 - 18490381 \n",
    "part2 = 18490381 - 36980761\n",
    "part3 = 36980761 - 55471141\n",
    "part4 = 55471141 - 73961521\n",
    "part5 = 73961521 - 92451902\n",
    "part6 = 92451901 - 110942282\n",
    "part7 = 110942282 - 129432662\n",
    "part8 = 129432662 - 147923042\n",
    "part9 = 147923042 - 166413422\n",
    "part10 = 166413422 -184903890"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.1"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}