Tabla 948 columnas count.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import pandas as pd\n",
    "from sklearn import preprocessing\n",
    "import category_encoders as ce\n",
    "\n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn.metrics import accuracy_score\n",
    "import numpy as np\n",
    "from sklearn.metrics import mean_squared_error\n",
    "from sklearn.ensemble import RandomForestClassifier\n",
    "import random\n",
    "import xgboost as xgb"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\Santi\\AppData\\Roaming\\Python\\Python36\\site-packages\\IPython\\core\\interactiveshell.py:2728: DtypeWarning: Columns (3,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22) have mixed types. Specify dtype option on import or set low_memory=False.\n",
      "  interactivity=interactivity, compiler=compiler, result=result)\n"
     ]
    }
   ],
   "source": [
    "datos = pd.read_csv('events_up_to_01062018.csv')\n",
    "tabla_completa = datos.copy()\n",
    "res = pd.read_csv('labels_training_set.csv');"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "tabla_completa['aux']=1\n",
    "#Voy a contar la cantidad para cada tipo de evento de cada persona\n",
    "num_eventos = tabla_completa.groupby(['person','event'])['aux'].sum()\n",
    "\n",
    "num_eventos = num_eventos.unstack().reset_index().fillna(0)\n",
    "num_eventos = num_eventos.set_index('person')\n",
    "num_eventos['total_eventos'] = num_eventos.sum(axis=1)\n",
    "#normalizo los numeros obtenidos según la cantidad de eventos totales de esa persona\n",
    "num_eventos_norm = num_eventos.loc[:,\"ad campaign hit\":\"visited site\"].div(num_eventos[\"total_eventos\"], axis=0)\n",
    "num_eventos_norm = num_eventos_norm.reset_index()\n",
    "num_eventos_norm.head()\n",
    "\n",
    "tabla_completa['timestamp'] = pd.to_datetime(tabla_completa['timestamp'])\n",
    "#tabla_completa['timestamp'].max() -tabla_completa['timestamp'].min()\n",
    "\n",
    "time_for_user = tabla_completa.groupby(['person'])['timestamp'].apply(lambda x: x.max() - x.min())\n",
    "time_for_user = time_for_user.reset_index()\n",
    "time_for_user['total_horas'] = time_for_user['timestamp'].dt.total_seconds()/3600\n",
    "time_for_user = time_for_user.drop(['timestamp'],axis='columns')\n",
    "\n",
    "max_time = tabla_completa.groupby(['person'])['timestamp'].apply(lambda x: x.max())\n",
    "max_time = max_time.reset_index()\n",
    "max_time['max'] = max_time['timestamp'].dt.month\n",
    "max_time = max_time.drop(['timestamp'],axis='columns')\n",
    "\n",
    "time_for_user = pd.merge(max_time,time_for_user,on='person',how='inner')\n",
    "\n",
    "min_time = tabla_completa.groupby(['person'])['timestamp'].apply(lambda x: x.min())\n",
    "min_time = min_time.reset_index()\n",
    "min_time['min'] = min_time['timestamp'].dt.month\n",
    "min_time = min_time.drop(['timestamp'],axis='columns')\n",
    "\n",
    "time_for_user = pd.merge(min_time,time_for_user,on='person',how='inner')\n",
    "\n",
    "num_eventos_norm = num_eventos_norm.reset_index()\n",
    "num_eventos = num_eventos.reset_index()\n",
    "total_eventos = pd.merge(num_eventos,num_eventos_norm, on='person',how='inner')\n",
    "total_eventos = pd.merge(total_eventos,time_for_user,on='person',how='inner')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(38829, 28)"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "total_eventos.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>timestamp</th>\n",
       "      <th>event</th>\n",
       "      <th>person</th>\n",
       "      <th>url</th>\n",
       "      <th>sku</th>\n",
       "      <th>model</th>\n",
       "      <th>condition</th>\n",
       "      <th>storage</th>\n",
       "      <th>color</th>\n",
       "      <th>skus</th>\n",
       "      <th>...</th>\n",
       "      <th>search_engine</th>\n",
       "      <th>channel</th>\n",
       "      <th>new_vs_returning</th>\n",
       "      <th>city</th>\n",
       "      <th>region</th>\n",
       "      <th>country</th>\n",
       "      <th>device_type</th>\n",
       "      <th>screen_resolution</th>\n",
       "      <th>operating_system_version</th>\n",
       "      <th>browser_version</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>2018-05-18 00:11:59</td>\n",
       "      <td>viewed product</td>\n",
       "      <td>4886f805</td>\n",
       "      <td>NaN</td>\n",
       "      <td>9288.0</td>\n",
       "      <td>Samsung Galaxy J7 Prime</td>\n",
       "      <td>Excelente</td>\n",
       "      <td>32GB</td>\n",
       "      <td>Dourado</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2018-05-18 00:11:27</td>\n",
       "      <td>viewed product</td>\n",
       "      <td>ad93850f</td>\n",
       "      <td>NaN</td>\n",
       "      <td>304.0</td>\n",
       "      <td>iPhone 5s</td>\n",
       "      <td>Muito Bom</td>\n",
       "      <td>32GB</td>\n",
       "      <td>Cinza espacial</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2018-05-18 00:11:16</td>\n",
       "      <td>viewed product</td>\n",
       "      <td>0297fc1e</td>\n",
       "      <td>NaN</td>\n",
       "      <td>6888.0</td>\n",
       "      <td>iPhone 6S</td>\n",
       "      <td>Muito Bom</td>\n",
       "      <td>64GB</td>\n",
       "      <td>Prateado</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>2018-05-18 00:11:14</td>\n",
       "      <td>viewed product</td>\n",
       "      <td>2d681dd8</td>\n",
       "      <td>NaN</td>\n",
       "      <td>11890.0</td>\n",
       "      <td>iPhone 7</td>\n",
       "      <td>Bom</td>\n",
       "      <td>128GB</td>\n",
       "      <td>Vermelho</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>2018-05-18 00:11:09</td>\n",
       "      <td>viewed product</td>\n",
       "      <td>cccea85e</td>\n",
       "      <td>NaN</td>\n",
       "      <td>7517.0</td>\n",
       "      <td>LG G4 H818P</td>\n",
       "      <td>Excelente</td>\n",
       "      <td>32GB</td>\n",
       "      <td>Branco</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 23 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "             timestamp           event    person  url      sku  \\\n",
       "0  2018-05-18 00:11:59  viewed product  4886f805  NaN   9288.0   \n",
       "1  2018-05-18 00:11:27  viewed product  ad93850f  NaN    304.0   \n",
       "2  2018-05-18 00:11:16  viewed product  0297fc1e  NaN   6888.0   \n",
       "3  2018-05-18 00:11:14  viewed product  2d681dd8  NaN  11890.0   \n",
       "4  2018-05-18 00:11:09  viewed product  cccea85e  NaN   7517.0   \n",
       "\n",
       "                     model  condition storage           color skus  \\\n",
       "0  Samsung Galaxy J7 Prime  Excelente    32GB         Dourado  NaN   \n",
       "1                iPhone 5s  Muito Bom    32GB  Cinza espacial  NaN   \n",
       "2                iPhone 6S  Muito Bom    64GB        Prateado  NaN   \n",
       "3                 iPhone 7        Bom   128GB        Vermelho  NaN   \n",
       "4              LG G4 H818P  Excelente    32GB          Branco  NaN   \n",
       "\n",
       "        ...       search_engine channel new_vs_returning city region country  \\\n",
       "0       ...                 NaN     NaN              NaN  NaN    NaN     NaN   \n",
       "1       ...                 NaN     NaN              NaN  NaN    NaN     NaN   \n",
       "2       ...                 NaN     NaN              NaN  NaN    NaN     NaN   \n",
       "3       ...                 NaN     NaN              NaN  NaN    NaN     NaN   \n",
       "4       ...                 NaN     NaN              NaN  NaN    NaN     NaN   \n",
       "\n",
       "  device_type screen_resolution operating_system_version browser_version  \n",
       "0         NaN               NaN                      NaN             NaN  \n",
       "1         NaN               NaN                      NaN             NaN  \n",
       "2         NaN               NaN                      NaN             NaN  \n",
       "3         NaN               NaN                      NaN             NaN  \n",
       "4         NaN               NaN                      NaN             NaN  \n",
       "\n",
       "[5 rows x 23 columns]"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "tabla_encode = datos.copy()\n",
    "tabla_encode.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "tabla_encode = tabla_encode.fillna(0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "device_type\n",
       "0             2137612\n",
       "Computer        97485\n",
       "Smartphone     103502\n",
       "Tablet           2799\n",
       "Unknown           283\n",
       "Name: person, dtype: int64"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#device_type\n",
    "tabla_encode.groupby('device_type')['person'].count()\n",
    "#0 no cuenta (es que no hay evento)\n",
    "#son solo 4 tipos distintos, así que los mantengo todos"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\Santi\\AppData\\Roaming\\Python\\Python36\\site-packages\\ipykernel_launcher.py:2: 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",
      "  \n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "(38829, 5)"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "dev_type = tabla_encode[['person','device_type']]\n",
    "dev_type['aux'] = 1\n",
    "dev_type = dev_type.groupby(['person','device_type'])['aux'].count().unstack().reset_index().fillna(0)\n",
    "dev_type = dev_type.drop([0],axis=1)\n",
    "dev_type = dev_type.set_index('person')\n",
    "dev_type = dev_type.add_prefix('device_type_')\n",
    "dev_type = dev_type.reset_index()\n",
    "dev_type.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "browser_version\n",
       "0                                  2137612\n",
       "Chrome 66.0                          57953\n",
       "Chrome Mobile 66.0                   45589\n",
       "Chrome Mobile 65.0                   15542\n",
       "Chrome 65.0                          14920\n",
       "Chrome Mobile 64.0                    7787\n",
       "Chrome 64.0                           5670\n",
       "Mobile Safari 11                      4501\n",
       "Chrome 63.0                           4358\n",
       "Chrome Mobile 63.0                    4217\n",
       "Samsung Internet 6.4                  2884\n",
       "Firefox 59                            2589\n",
       "Firefox 60                            2298\n",
       "Chrome Mobile 39                      2042\n",
       "Mobile Safari 10                      1970\n",
       "Edge 16.16299                         1574\n",
       "Chrome Mobile 55.0                    1549\n",
       "Chrome Mobile 43.0                    1266\n",
       "Chrome 49.0                           1028\n",
       "IE 11                                  999\n",
       "Chrome Mobile 62.0                     891\n",
       "Samsung Internet 3.3                   820\n",
       "Chrome 67.0                            798\n",
       "Mobile Safari 9                        749\n",
       "Opera 52.0                             720\n",
       "Chrome Mobile 61.0                     708\n",
       "Chrome 58.0                            697\n",
       "Chrome Mobile 56.0                     695\n",
       "Facebook 173                           664\n",
       "Facebook 172                           561\n",
       "                                    ...   \n",
       "BingPreview 1                            1\n",
       "Opera Mobile 28.0                        1\n",
       "Chrome 35.0                              1\n",
       "Chrome 53.19                             1\n",
       "Opera Mini 35.0                          1\n",
       "Opera Mini 31.0                          1\n",
       "Chromium 64.0                            1\n",
       "Facebook 139                             1\n",
       "Facebook 145                             1\n",
       "Facebook 147                             1\n",
       "Facebook 151                             1\n",
       "Facebook 68                              1\n",
       "Firefox 15                               1\n",
       "Firefox 15.0                             1\n",
       "Chrome 57.7                              1\n",
       "Chrome 57.3                              1\n",
       "Chrome 57.21                             1\n",
       "Firefox 24                               1\n",
       "Firefox 26                               1\n",
       "Firefox 28                               1\n",
       "K-Meleon 76                              1\n",
       "Chrome 56.13                             1\n",
       "Mobile Safari 9.2                        1\n",
       "Mobile Safari UI/WKWebView 10.2          1\n",
       "Chrome 48.4                              1\n",
       "Mobile Safari UI/WKWebView 11.3          1\n",
       "Opera 42.0                               1\n",
       "Opera 47.0                               1\n",
       "Opera Mini 28.0                          1\n",
       "Samsung Internet 1.1                     1\n",
       "Name: person, Length: 367, dtype: int64"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#device_type\n",
    "tabla_encode.groupby('browser_version')['person'].count().sort_values(ascending=False)\n",
    "#0 no cuenta (es que no hay evento)\n",
    "#son 367 tipos distintos. Voy a vovler a mirar, pero uniendo por versiones"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "3               Chrome\n",
       "4        Chrome Mobile\n",
       "16       Mobile Safari\n",
       "10             Firefox\n",
       "25    Samsung Internet\n",
       "Name: browser_version, dtype: object"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "test_b = tabla_encode['browser_version'].str.replace('\\d+', '')\n",
    "test_b = test_b.str.replace('.', '')\n",
    "test_b = test_b.str.strip()\n",
    "test_b2 = pd.DataFrame(test_b)\n",
    "test_b2['aux']=1\n",
    "test_b2 = test_b2.groupby('browser_version').count().reset_index().sort_values(by='aux',ascending=False)\n",
    "#Ahora me quedan 30 browsers. Voy a quedarme con los 5 más usados. El resto irá a una columna \"otros\".\n",
    "browsers = test_b2['browser_version'].head()\n",
    "browsers"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\Santi\\AppData\\Roaming\\Python\\Python36\\site-packages\\ipykernel_launcher.py:2: 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",
      "  \n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "(38829, 367)"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "browser_version = tabla_encode[['person','browser_version']]\n",
    "browser_version['aux'] = 1\n",
    "browser_version = browser_version.groupby(['person','browser_version'])['aux'].count().unstack().reset_index().fillna(0)\n",
    "browser_version = browser_version.drop([0],axis=1)\n",
    "browser_version = browser_version.set_index('person')\n",
    "browser_version = browser_version.add_prefix('browser_version_')\n",
    "browser_version = browser_version.reset_index()\n",
    "browser_version.shape\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "241      360x640\n",
       "83      1366x768\n",
       "150    1920x1080\n",
       "220      320x534\n",
       "79      1360x768\n",
       "Name: screen_resolution, dtype: object"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#screen_resolution\n",
    "test_s = tabla_encode['screen_resolution']\n",
    "test_s = pd.DataFrame(test_s)\n",
    "test_s['aux']=1\n",
    "test_s = test_s.groupby('screen_resolution').count().reset_index().sort_values(by='aux',ascending=False)\n",
    "test_s = test_s.drop([0],axis=0)\n",
    "\n",
    "resolutions = test_s['screen_resolution'].head()\n",
    "resolutions\n",
    "#Me quedo con las 5 resoluciones que mas buscan"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\Santi\\AppData\\Roaming\\Python\\Python36\\site-packages\\ipykernel_launcher.py:2: 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",
      "  \n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "(38829, 394)"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "screen_resolution = tabla_encode[['person','screen_resolution']]\n",
    "screen_resolution['aux'] = 1\n",
    "screen_resolution = screen_resolution.groupby(['person','screen_resolution'])['aux'].count().unstack().reset_index().fillna(0)\n",
    "screen_resolution = screen_resolution.drop([0],axis=1)\n",
    "screen_resolution = screen_resolution.set_index('person')\n",
    "screen_resolution = screen_resolution.add_prefix('screen_resolution_')\n",
    "screen_resolution = screen_resolution.reset_index()\n",
    "screen_resolution.shape\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0           Android\n",
       "11          Windows\n",
       "16              iOS\n",
       "15       Windows XP\n",
       "12    Windows Phone\n",
       "Name: operating_system_version, dtype: object"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#operating_system_version\n",
    "test_os = tabla_encode['operating_system_version'].str.replace('\\d+', '')\n",
    "test_os = test_os.str.replace('.', '')\n",
    "test_os = test_os.str.strip()\n",
    "test_os2 = pd.DataFrame(test_os)\n",
    "test_os2['aux']=1\n",
    "test_os2 = test_os2.groupby('operating_system_version').count().reset_index().sort_values(by='aux',ascending=False)\n",
    "test_os2\n",
    "\n",
    "OSs = test_os2['operating_system_version'].head()\n",
    "OSs\n",
    "#Me quedo con las 5 resoluciones que mas buscan"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\Santi\\AppData\\Roaming\\Python\\Python36\\site-packages\\ipykernel_launcher.py:2: 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",
      "  \n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "(38829, 132)"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "operating_system_version = tabla_encode[['person','operating_system_version']]\n",
    "operating_system_version['aux'] = 1\n",
    "operating_system_version = operating_system_version.groupby(['person','operating_system_version'])['aux'].count().unstack().reset_index().fillna(0)\n",
    "operating_system_version = operating_system_version.drop([0],axis=1)\n",
    "operating_system_version = operating_system_version.set_index('person')\n",
    "operating_system_version = operating_system_version.add_prefix('operating_system_version_')\n",
    "operating_system_version = operating_system_version.reset_index()\n",
    "operating_system_version.shape\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "2     16GB\n",
       "4     32GB\n",
       "7     64GB\n",
       "1    128GB\n",
       "8      8GB\n",
       "Name: storage, dtype: object"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#storage\n",
    "test_st = tabla_encode['storage']\n",
    "test_st = pd.DataFrame(test_st)\n",
    "test_st['aux']=1\n",
    "test_st = test_st.groupby('storage').count().reset_index().sort_values(by='aux',ascending=False)\n",
    "test_st = test_st.drop([0],axis=0)\n",
    "\n",
    "storages = test_st['storage'].head()\n",
    "storages\n",
    "#Me quedo con las 5 resoluciones que mas buscan"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\Santi\\AppData\\Roaming\\Python\\Python36\\site-packages\\ipykernel_launcher.py:2: 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",
      "  \n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "(38829, 9)"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "storage = tabla_encode[['person','storage']]\n",
    "storage['aux'] = 1\n",
    "storage = storage.groupby(['person','storage'])['aux'].count().unstack().reset_index().fillna(0)\n",
    "storage = storage.drop([0],axis=1)\n",
    "storage = storage.set_index('person')\n",
    "storage = storage.add_prefix('storage_')\n",
    "storage = storage.reset_index()\n",
    "storage.shape\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>condition</th>\n",
       "      <th>aux</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>Bom</td>\n",
       "      <td>547617</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>Excelente</td>\n",
       "      <td>378697</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>Muito Bom</td>\n",
       "      <td>359053</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>Bom - Sem Touch ID</td>\n",
       "      <td>31742</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>Novo</td>\n",
       "      <td>3421</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "            condition     aux\n",
       "1                 Bom  547617\n",
       "3           Excelente  378697\n",
       "4           Muito Bom  359053\n",
       "2  Bom - Sem Touch ID   31742\n",
       "5                Novo    3421"
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#condition\n",
    "test_c = tabla_encode['condition']\n",
    "test_c = pd.DataFrame(test_c)\n",
    "test_c['aux']=1\n",
    "test_c = test_c.groupby('condition').count().reset_index().sort_values(by='aux',ascending=False)\n",
    "test_c = test_c.drop([0],axis=0)\n",
    "test_c\n",
    "\n",
    "#Ya son 5, me quedo todas."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\Santi\\AppData\\Roaming\\Python\\Python36\\site-packages\\ipykernel_launcher.py:2: 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",
      "  \n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "(38829, 6)"
      ]
     },
     "execution_count": 19,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "condition = tabla_encode[['person','condition']]\n",
    "condition['aux'] = 1\n",
    "condition = condition.groupby(['person','condition'])['aux'].count().unstack().reset_index().fillna(0)\n",
    "condition = condition.drop([0],axis=1)\n",
    "condition = condition.set_index('person')\n",
    "condition = condition.add_prefix('condition_')\n",
    "condition = condition.reset_index()\n",
    "condition.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>new_vs_returning</th>\n",
       "      <th>aux</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>Returning</td>\n",
       "      <td>165827</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>New</td>\n",
       "      <td>38242</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "  new_vs_returning     aux\n",
       "2        Returning  165827\n",
       "1              New   38242"
      ]
     },
     "execution_count": 20,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#new_vs_returning\n",
    "test_nv = tabla_encode['new_vs_returning']\n",
    "test_nv = pd.DataFrame(test_nv)\n",
    "test_nv['aux']=1\n",
    "test_nv = test_nv.groupby('new_vs_returning').count().reset_index().sort_values(by='aux',ascending=False)\n",
    "test_nv = test_nv.drop([0],axis=0)\n",
    "test_nv\n",
    "#Son solo 2 valores"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\Santi\\AppData\\Roaming\\Python\\Python36\\site-packages\\ipykernel_launcher.py:3: 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",
      "  This is separate from the ipykernel package so we can avoid doing imports until\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "(38829, 3)"
      ]
     },
     "execution_count": 21,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#Esta la hago un count.\n",
    "new_vs_returning = tabla_encode[['person','new_vs_returning']]\n",
    "new_vs_returning['aux'] = 1\n",
    "new_vs_returning = new_vs_returning.groupby(['person','new_vs_returning'])['aux'].count().unstack().reset_index().fillna(0)\n",
    "new_vs_returning = new_vs_returning.drop([0],axis=1)\n",
    "new_vs_returning = new_vs_returning.set_index('person')\n",
    "new_vs_returning = new_vs_returning.add_prefix('new_vs_returning_')\n",
    "new_vs_returning = new_vs_returning.reset_index()\n",
    "new_vs_returning.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>search_engine</th>\n",
       "      <th>aux</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>Google</td>\n",
       "      <td>105195</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>Bing</td>\n",
       "      <td>676</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>Yahoo</td>\n",
       "      <td>519</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>Ask</td>\n",
       "      <td>16</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "  search_engine     aux\n",
       "3        Google  105195\n",
       "2          Bing     676\n",
       "4         Yahoo     519\n",
       "1           Ask      16"
      ]
     },
     "execution_count": 22,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#search_engine\n",
    "test_se = tabla_encode['search_engine']\n",
    "test_se = pd.DataFrame(test_se)\n",
    "test_se['aux']=1\n",
    "test_se = test_se.groupby('search_engine').count().reset_index().sort_values(by='aux',ascending=False)\n",
    "test_se = test_se.drop([0],axis=0)\n",
    "test_se\n",
    "#Son solo 4 valores"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\Santi\\AppData\\Roaming\\Python\\Python36\\site-packages\\ipykernel_launcher.py:3: 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",
      "  This is separate from the ipykernel package so we can avoid doing imports until\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "(38829, 5)"
      ]
     },
     "execution_count": 23,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#Esta la hago un count.\n",
    "search_engine = tabla_encode[['person','search_engine']]\n",
    "search_engine['aux'] = 1\n",
    "search_engine = search_engine.groupby(['person','search_engine'])['aux'].count().unstack().reset_index().fillna(0)\n",
    "search_engine = search_engine.drop([0],axis=1)\n",
    "search_engine = search_engine.set_index('person')\n",
    "search_engine = search_engine.add_prefix('search_engine_')\n",
    "search_engine = search_engine.reset_index()\n",
    "search_engine.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>channel</th>\n",
       "      <th>aux</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>Paid</td>\n",
       "      <td>91753</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>Direct</td>\n",
       "      <td>50675</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>Referral</td>\n",
       "      <td>33990</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>Organic</td>\n",
       "      <td>25867</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>Social</td>\n",
       "      <td>1406</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>Email</td>\n",
       "      <td>371</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>Unknown</td>\n",
       "      <td>7</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "    channel    aux\n",
       "4      Paid  91753\n",
       "1    Direct  50675\n",
       "5  Referral  33990\n",
       "3   Organic  25867\n",
       "6    Social   1406\n",
       "2     Email    371\n",
       "7   Unknown      7"
      ]
     },
     "execution_count": 24,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#channel\n",
    "test_ch = tabla_encode['channel']\n",
    "test_ch = pd.DataFrame(test_ch)\n",
    "test_ch['aux']=1\n",
    "test_ch = test_ch.groupby('channel').count().reset_index().sort_values(by='aux',ascending=False)\n",
    "test_ch = test_ch.drop([0],axis=0)\n",
    "test_ch\n",
    "#Mantengo los 7 valores"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\Santi\\AppData\\Roaming\\Python\\Python36\\site-packages\\ipykernel_launcher.py:3: 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",
      "  This is separate from the ipykernel package so we can avoid doing imports until\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "(38829, 8)"
      ]
     },
     "execution_count": 25,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#Esta la hago un count.\n",
    "channel = tabla_encode[['person','channel']]\n",
    "channel['aux'] = 1\n",
    "channel = channel.groupby(['person','channel'])['aux'].count().unstack().reset_index().fillna(0)\n",
    "channel = channel.drop([0],axis=1)\n",
    "channel = channel.set_index('person')\n",
    "channel = channel.add_prefix('channel_')\n",
    "channel = channel.reset_index()\n",
    "channel.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [],
   "source": [
    "#Ahora a unir todo..."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(38829, 948)"
      ]
     },
     "execution_count": 27,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "tabla_final = pd.merge(total_eventos,dev_type,on='person',how='inner')\n",
    "tabla_final = pd.merge(tabla_final,screen_resolution,on='person',how='inner')\n",
    "tabla_final = pd.merge(tabla_final,storage,on='person',how='inner')\n",
    "tabla_final = pd.merge(tabla_final,condition,on='person',how='inner')\n",
    "tabla_final = pd.merge(tabla_final,new_vs_returning,on='person',how='inner')\n",
    "tabla_final = pd.merge(tabla_final,search_engine,on='person',how='inner')\n",
    "tabla_final = pd.merge(tabla_final,channel,on='person',how='inner')\n",
    "tabla_final = pd.merge(tabla_final,browser_version,on='person',how='left')\n",
    "tabla_final = pd.merge(tabla_final,operating_system_version,on='person',how='left')\n",
    "tabla_final = tabla_final.fillna(0)\n",
    "tabla_final.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "Int64Index: 38829 entries, 0 to 38828\n",
      "Columns: 948 entries, person to operating_system_version_iOS 9.3.5\n",
      "dtypes: float64(944), int64(3), object(1)\n",
      "memory usage: 281.1+ MB\n"
     ]
    }
   ],
   "source": [
    "tabla_final.info()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [],
   "source": [
    "\n",
    "intento_f = pd.merge(tabla_final,res,on='person',how='inner')\n",
    "\n",
    "intento_f = intento_f.set_index('person')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [],
   "source": [
    "final = pd.read_csv('trocafone_kaggle_test.csv')\n",
    "final = pd.merge(tabla_final,final,on='person',how='inner')\n",
    "final = final.set_index('person')\n",
    "x_entrega = final"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [],
   "source": [
    "#XGBoost\n",
    "\n",
    "X, y = intento_f.iloc[:,:-1],intento_f.iloc[:,-1]\n",
    "data_dmatrix = xgb.DMatrix(data=X,label=y)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([0.025     , 0.01666667, 0.02083333, 0.02916667, 0.00833333,\n",
       "       0.        , 0.01666667, 0.00833333, 0.00416667, 0.02083333,\n",
       "       0.00833333, 0.03333334, 0.04583333, 0.01666667, 0.00833333,\n",
       "       0.025     , 0.03333334, 0.        , 0.        , 0.00833333,\n",
       "       0.00416667, 0.00416667, 0.05416667, 0.01666667, 0.01666667,\n",
       "       0.00833333, 0.03333334, 0.03333334, 0.00833333, 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.00416667, 0.        ,\n",
       "       0.        , 0.00833333, 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.025     , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.00416667,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.00416667, 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.00416667, 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.00416667,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.01666667, 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.02083333, 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.0125    , 0.        , 0.00416667, 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.00416667, 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.0125    ,\n",
       "       0.02916667, 0.        , 0.00416667, 0.        , 0.00416667,\n",
       "       0.00416667, 0.0375    , 0.00416667, 0.00833333, 0.00833333,\n",
       "       0.00833333, 0.        , 0.        , 0.00416667, 0.        ,\n",
       "       0.00416667, 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.0125    , 0.        , 0.        , 0.00416667, 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.00416667, 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.00833333, 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.00416667, 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.00833333,\n",
       "       0.        , 0.0125    , 0.00833333, 0.00416667, 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.00416667, 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.00833333,\n",
       "       0.        , 0.00833333, 0.00416667, 0.00416667, 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.00416667, 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.00416667, 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.00416667,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.00416667, 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.00416667, 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.01666667,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.00833333, 0.        , 0.        , 0.00416667,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.00416667, 0.00833333, 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.00416667, 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.00416667, 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.00416667, 0.00416667, 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.025     , 0.        , 0.00416667, 0.00416667, 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.0125    ,\n",
       "       0.00833333, 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.00416667, 0.        , 0.00833333, 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.00416667, 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.00416667, 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        , 0.        , 0.        , 0.        ,\n",
       "       0.        , 0.        ], dtype=float32)"
      ]
     },
     "execution_count": 32,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from sklearn.model_selection import train_test_split\n",
    "\n",
    "X_train, X_test, y_train, y_test = \\\n",
    "    train_test_split(X, y, test_size=0.2, random_state=123)\n",
    "    \n",
    "xg_reg = xgb.XGBRegressor(objective ='reg:linear', \n",
    "                colsample_bytree = 0.3, learning_rate = 0.1,\n",
    "                max_depth = 5, alpha = 2, n_estimators = 10)\n",
    "\n",
    "model = xg_reg.fit(X_train,y_train)\n",
    "model.feature_importances_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [
    {
     "ename": "NameError",
     "evalue": "name 'SelectFromModel' is not defined",
     "output_type": "error",
     "traceback": [
      "\u001b[1;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[1;31mNameError\u001b[0m                                 Traceback (most recent call last)",
      "\u001b[1;32m<ipython-input-33-a23b6f0ede16>\u001b[0m in \u001b[0;36m<module>\u001b[1;34m()\u001b[0m\n\u001b[1;32m----> 1\u001b[1;33m \u001b[0mselection\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mSelectFromModel\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mmodel\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mthreshold\u001b[0m\u001b[1;33m=\u001b[0m\u001b[1;36m0\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mprefit\u001b[0m\u001b[1;33m=\u001b[0m\u001b[1;32mTrue\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m      2\u001b[0m \u001b[0mselect_X_train\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mselection\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mtransform\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mX_train\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m      3\u001b[0m \u001b[0mselect_X_test\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mselection\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mtransform\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mX_test\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m      4\u001b[0m \u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m      5\u001b[0m \u001b[0mxg_reg\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mfit\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mselect_X_train\u001b[0m\u001b[1;33m,\u001b[0m\u001b[0my_train\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;31mNameError\u001b[0m: name 'SelectFromModel' is not defined"
     ]
    }
   ],
   "source": [
    "selection = SelectFromModel(model, threshold=0, prefit=True)\n",
    "select_X_train = selection.transform(X_train)\n",
    "select_X_test = selection.transform(X_test)\n",
    "\n",
    "xg_reg.fit(select_X_train,y_train)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<function matplotlib.pyplot.show>"
      ]
     },
     "execution_count": 39,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# plot feature importance using built-in function\n",
    "from numpy import loadtxt\n",
    "from xgboost import XGBClassifier\n",
    "from xgboost import plot_importance\n",
    "from matplotlib import pyplot\n",
    "\n",
    "# plot feature importance\n",
    "plot_importance(model)\n",
    "pyplot.show"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "from numpy import loadtxt\n",
    "from numpy import sort\n",
    "from xgboost import XGBClassifier\n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn.metrics import accuracy_score\n",
    "from sklearn.feature_selection import SelectFromModel\n",
    "# load data\n",
    "\n",
    "# make predictions for test data and evaluate\n",
    "y_pred = model.predict(X_test)\n",
    "predictions = [round(value) for value in y_pred]\n",
    "accuracy = accuracy_score(y_test, predictions)\n",
    "print(\"Accuracy: %.2f%%\" % (accuracy * 100.0))\n",
    "# Fit model using each importance as a threshold\n",
    "thresholds = [0,1e-12,1e-11,1e-10,1e-9,1e-8,1e-7,1e-6,1e-5,1e-4,1e-3,1e-2,1e-1]\n",
    "for thresh in thresholds:\n",
    "    # select features using threshold\n",
    "    selection = SelectFromModel(model, threshold=thresh, prefit=True)\n",
    "    select_X_train = selection.transform(X_train)\n",
    "    # train model\n",
    "    selection_model = XGBClassifier()\n",
    "    selection_model.fit(select_X_train, y_train)\n",
    "    # eval model\n",
    "    select_X_test = selection.transform(X_test)\n",
    "    y_pred = selection_model.predict(select_X_test)\n",
    "    predictions = [round(value) for value in y_pred]\n",
    "    accuracy = accuracy_score(y_test, predictions)\n",
    "    print(\"Thresh=%.3f, n=%d, Accuracy: %.2f%%\" % (thresh, select_X_train.shape[1], accuracy*100.0))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "preds = xg_reg.predict(select_X_test)\n",
    "rmse = np.sqrt(mean_squared_error(y_test, preds))\n",
    "print(\"RMSE: %f\" % (rmse))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "\n",
    "entrega = xg_reg.predict(x_entrega)\n",
    "entrega = entrega.clip(min=0)  \n",
    "\n",
    "y_summit = pd.DataFrame(entrega)\n",
    "y_summit.columns = ['label']\n",
    "y_summit['person'] = x_entrega.reset_index()['person']\n",
    "y_person = x_entrega.reset_index()['person']\n",
    "\n",
    "y_summit = y_summit[['person','label']]\n",
    "y_summit.to_csv('summit028.csv', index=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.4"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}