Anls_OLS_validation.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Data Analysis: Validate OLS regression model\n",
    "\n",
    "## This notebook:\n",
    "1. Splits training and testing datasets\n",
    "2. Validates the optimal model from ('Anls_OLS_weather')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Populating the interactive namespace from numpy and matplotlib\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/Yuwen/anaconda/lib/python3.6/site-packages/IPython/core/magics/pylab.py:161: UserWarning: pylab import has clobbered these variables: ['test']\n",
      "`%matplotlib` prevents importing * from pylab and numpy\n",
      "  \"\\n`%matplotlib` prevents importing * from pylab and numpy\"\n"
     ]
    }
   ],
   "source": [
    "import pylab as pl\n",
    "import pandas as pd\n",
    "import statsmodels.formula.api as smf\n",
    "from sklearn.model_selection import train_test_split\n",
    "\n",
    "%pylab inline"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Split Train and Test datasets"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "d = pd.read_csv('merged3_final.csv')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style>\n",
       "    .dataframe thead tr:only-child th {\n",
       "        text-align: right;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: left;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Unnamed: 0</th>\n",
       "      <th>tripduration</th>\n",
       "      <th>starttime</th>\n",
       "      <th>stoptime</th>\n",
       "      <th>start station id</th>\n",
       "      <th>start station name</th>\n",
       "      <th>start station latitude</th>\n",
       "      <th>start station longitude</th>\n",
       "      <th>end station id</th>\n",
       "      <th>end station name</th>\n",
       "      <th>...</th>\n",
       "      <th>usertype</th>\n",
       "      <th>birth year</th>\n",
       "      <th>gender</th>\n",
       "      <th>startdate</th>\n",
       "      <th>stopdate</th>\n",
       "      <th>Event_type</th>\n",
       "      <th>End_Time</th>\n",
       "      <th>weekday</th>\n",
       "      <th>O_date</th>\n",
       "      <th>O_hour</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0</td>\n",
       "      <td>173</td>\n",
       "      <td>16:03:10</td>\n",
       "      <td>16:06:03</td>\n",
       "      <td>243</td>\n",
       "      <td>Fulton St &amp; Rockwell Pl</td>\n",
       "      <td>40.688226</td>\n",
       "      <td>-73.979382</td>\n",
       "      <td>241</td>\n",
       "      <td>DeKalb Ave &amp; S Portland Ave</td>\n",
       "      <td>...</td>\n",
       "      <td>Subscriber</td>\n",
       "      <td>1971.0</td>\n",
       "      <td>2</td>\n",
       "      <td>2016-01-01</td>\n",
       "      <td>2016-01-01</td>\n",
       "      <td>no-event</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>2016-01-01</td>\n",
       "      <td>16</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1</td>\n",
       "      <td>136</td>\n",
       "      <td>16:05:54</td>\n",
       "      <td>16:08:11</td>\n",
       "      <td>420</td>\n",
       "      <td>Clermont Ave &amp; Lafayette Ave</td>\n",
       "      <td>40.687645</td>\n",
       "      <td>-73.969689</td>\n",
       "      <td>270</td>\n",
       "      <td>Adelphi St &amp; Myrtle Ave</td>\n",
       "      <td>...</td>\n",
       "      <td>Subscriber</td>\n",
       "      <td>1980.0</td>\n",
       "      <td>1</td>\n",
       "      <td>2016-01-01</td>\n",
       "      <td>2016-01-01</td>\n",
       "      <td>no-event</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>2016-01-01</td>\n",
       "      <td>16</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2</td>\n",
       "      <td>653</td>\n",
       "      <td>16:13:47</td>\n",
       "      <td>16:24:40</td>\n",
       "      <td>83</td>\n",
       "      <td>Atlantic Ave &amp; Fort Greene Pl</td>\n",
       "      <td>40.683826</td>\n",
       "      <td>-73.976323</td>\n",
       "      <td>278</td>\n",
       "      <td>Concord St &amp; Bridge St</td>\n",
       "      <td>...</td>\n",
       "      <td>Subscriber</td>\n",
       "      <td>1976.0</td>\n",
       "      <td>1</td>\n",
       "      <td>2016-01-01</td>\n",
       "      <td>2016-01-01</td>\n",
       "      <td>no-event</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>2016-01-01</td>\n",
       "      <td>16</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>3</td>\n",
       "      <td>659</td>\n",
       "      <td>16:13:47</td>\n",
       "      <td>16:24:46</td>\n",
       "      <td>83</td>\n",
       "      <td>Atlantic Ave &amp; Fort Greene Pl</td>\n",
       "      <td>40.683826</td>\n",
       "      <td>-73.976323</td>\n",
       "      <td>278</td>\n",
       "      <td>Concord St &amp; Bridge St</td>\n",
       "      <td>...</td>\n",
       "      <td>Subscriber</td>\n",
       "      <td>1985.0</td>\n",
       "      <td>2</td>\n",
       "      <td>2016-01-01</td>\n",
       "      <td>2016-01-01</td>\n",
       "      <td>no-event</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>2016-01-01</td>\n",
       "      <td>16</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>4</td>\n",
       "      <td>1419</td>\n",
       "      <td>16:20:39</td>\n",
       "      <td>16:44:19</td>\n",
       "      <td>83</td>\n",
       "      <td>Atlantic Ave &amp; Fort Greene Pl</td>\n",
       "      <td>40.683826</td>\n",
       "      <td>-73.976323</td>\n",
       "      <td>532</td>\n",
       "      <td>S 5 Pl &amp; S 4 St</td>\n",
       "      <td>...</td>\n",
       "      <td>Subscriber</td>\n",
       "      <td>1993.0</td>\n",
       "      <td>1</td>\n",
       "      <td>2016-01-01</td>\n",
       "      <td>2016-01-01</td>\n",
       "      <td>no-event</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>2016-01-01</td>\n",
       "      <td>16</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 23 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "   Unnamed: 0  tripduration starttime  stoptime  start station id  \\\n",
       "0           0           173  16:03:10  16:06:03               243   \n",
       "1           1           136  16:05:54  16:08:11               420   \n",
       "2           2           653  16:13:47  16:24:40                83   \n",
       "3           3           659  16:13:47  16:24:46                83   \n",
       "4           4          1419  16:20:39  16:44:19                83   \n",
       "\n",
       "              start station name  start station latitude  \\\n",
       "0        Fulton St & Rockwell Pl               40.688226   \n",
       "1   Clermont Ave & Lafayette Ave               40.687645   \n",
       "2  Atlantic Ave & Fort Greene Pl               40.683826   \n",
       "3  Atlantic Ave & Fort Greene Pl               40.683826   \n",
       "4  Atlantic Ave & Fort Greene Pl               40.683826   \n",
       "\n",
       "   start station longitude  end station id             end station name  \\\n",
       "0               -73.979382             241  DeKalb Ave & S Portland Ave   \n",
       "1               -73.969689             270      Adelphi St & Myrtle Ave   \n",
       "2               -73.976323             278       Concord St & Bridge St   \n",
       "3               -73.976323             278       Concord St & Bridge St   \n",
       "4               -73.976323             532              S 5 Pl & S 4 St   \n",
       "\n",
       "   ...      usertype  birth year  gender   startdate    stopdate  Event_type  \\\n",
       "0  ...    Subscriber      1971.0       2  2016-01-01  2016-01-01    no-event   \n",
       "1  ...    Subscriber      1980.0       1  2016-01-01  2016-01-01    no-event   \n",
       "2  ...    Subscriber      1976.0       1  2016-01-01  2016-01-01    no-event   \n",
       "3  ...    Subscriber      1985.0       2  2016-01-01  2016-01-01    no-event   \n",
       "4  ...    Subscriber      1993.0       1  2016-01-01  2016-01-01    no-event   \n",
       "\n",
       "  End_Time weekday      O_date O_hour  \n",
       "0      NaN     NaN  2016-01-01     16  \n",
       "1      NaN     NaN  2016-01-01     16  \n",
       "2      NaN     NaN  2016-01-01     16  \n",
       "3      NaN     NaN  2016-01-01     16  \n",
       "4      NaN     NaN  2016-01-01     16  \n",
       "\n",
       "[5 rows x 23 columns]"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data = pd.read_csv('merged2_event_CB2016_800m.csv')\n",
    "data.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "60148"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(data)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "stdate = data.O_date.unique()\n",
    "sthour = data.O_hour.unique()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "train = pd.DataFrame()\n",
    "test = pd.DataFrame()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "for i in stdate:\n",
    "    for j in sthour:\n",
    "        X_train, X_test = train_test_split(data[(data['O_date']==i)&(data['O_hour']==j)], \n",
    "                                           test_size = 0.3, random_state = 200)\n",
    "        train = pd.concat([train,X_train])\n",
    "        test = pd.concat([test,X_test])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "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>Unnamed: 0</th>\n",
       "      <th>O_date</th>\n",
       "      <th>Event_type</th>\n",
       "      <th>O_hour</th>\n",
       "      <th>Count</th>\n",
       "      <th>date</th>\n",
       "      <th>temperature</th>\n",
       "      <th>precipitation</th>\n",
       "      <th>weekday</th>\n",
       "      <th>O_month</th>\n",
       "      <th>season</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0</td>\n",
       "      <td>2016-01-01</td>\n",
       "      <td>no-event</td>\n",
       "      <td>16</td>\n",
       "      <td>8</td>\n",
       "      <td>2016-01-01</td>\n",
       "      <td>38.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>True</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1</td>\n",
       "      <td>2016-01-01</td>\n",
       "      <td>no-event</td>\n",
       "      <td>17</td>\n",
       "      <td>7</td>\n",
       "      <td>2016-01-01</td>\n",
       "      <td>38.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>True</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2</td>\n",
       "      <td>2016-01-01</td>\n",
       "      <td>no-event</td>\n",
       "      <td>18</td>\n",
       "      <td>4</td>\n",
       "      <td>2016-01-01</td>\n",
       "      <td>38.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>True</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>3</td>\n",
       "      <td>2016-01-01</td>\n",
       "      <td>no-event</td>\n",
       "      <td>19</td>\n",
       "      <td>2</td>\n",
       "      <td>2016-01-01</td>\n",
       "      <td>38.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>True</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>4</td>\n",
       "      <td>2016-01-01</td>\n",
       "      <td>no-event</td>\n",
       "      <td>20</td>\n",
       "      <td>2</td>\n",
       "      <td>2016-01-01</td>\n",
       "      <td>38.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>True</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   Unnamed: 0      O_date Event_type  O_hour  Count        date  temperature  \\\n",
       "0           0  2016-01-01   no-event      16      8  2016-01-01         38.0   \n",
       "1           1  2016-01-01   no-event      17      7  2016-01-01         38.0   \n",
       "2           2  2016-01-01   no-event      18      4  2016-01-01         38.0   \n",
       "3           3  2016-01-01   no-event      19      2  2016-01-01         38.0   \n",
       "4           4  2016-01-01   no-event      20      2  2016-01-01         38.0   \n",
       "\n",
       "   precipitation  weekday  O_month  season  \n",
       "0            0.0     True        1       1  \n",
       "1            0.0     True        1       1  \n",
       "2            0.0     True        1       1  \n",
       "3            0.0     True        1       1  \n",
       "4            0.0     True        1       1  "
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "d.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "train, test = train_test_split(d, test_size = 0.3, random_state = 200)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(2015, 864, 2879)"
      ]
     },
     "execution_count": 37,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(train), len(test), len(train) + len(test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "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>Unnamed: 0</th>\n",
       "      <th>O_date</th>\n",
       "      <th>Event_type</th>\n",
       "      <th>O_hour</th>\n",
       "      <th>Count</th>\n",
       "      <th>date</th>\n",
       "      <th>temperature</th>\n",
       "      <th>precipitation</th>\n",
       "      <th>weekday</th>\n",
       "      <th>O_month</th>\n",
       "      <th>season</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>305</th>\n",
       "      <td>305</td>\n",
       "      <td>2016-02-12</td>\n",
       "      <td>other</td>\n",
       "      <td>23</td>\n",
       "      <td>5</td>\n",
       "      <td>2016-02-12</td>\n",
       "      <td>45.5</td>\n",
       "      <td>0.0</td>\n",
       "      <td>True</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "     Unnamed: 0      O_date Event_type  O_hour  Count        date  \\\n",
       "305         305  2016-02-12      other      23      5  2016-02-12   \n",
       "\n",
       "     temperature  precipitation  weekday  O_month  season  \n",
       "305         45.5            0.0     True        2       1  "
      ]
     },
     "execution_count": 38,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train.head(1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "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>O_date</th>\n",
       "      <th>Event_type</th>\n",
       "      <th>O_hour</th>\n",
       "      <th>Count</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>2016-01-01</td>\n",
       "      <td>no-event</td>\n",
       "      <td>16</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2016-01-01</td>\n",
       "      <td>no-event</td>\n",
       "      <td>18</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2016-01-01</td>\n",
       "      <td>no-event</td>\n",
       "      <td>21</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>2016-01-01</td>\n",
       "      <td>no-event</td>\n",
       "      <td>22</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>2016-01-01</td>\n",
       "      <td>no-event</td>\n",
       "      <td>23</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       O_date Event_type  O_hour  Count\n",
       "0  2016-01-01   no-event      16      1\n",
       "1  2016-01-01   no-event      18      1\n",
       "2  2016-01-01   no-event      21      1\n",
       "3  2016-01-01   no-event      22      1\n",
       "4  2016-01-01   no-event      23      1"
      ]
     },
     "execution_count": 39,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "d_train = pd.DataFrame(train.groupby(['O_date','Event_type','O_hour'],as_index=False).size())\n",
    "d_train.reset_index(inplace=True)\n",
    "d_train.columns = ['O_date','Event_type','O_hour','Count']\n",
    "d_train.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "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>O_date</th>\n",
       "      <th>Event_type</th>\n",
       "      <th>O_hour</th>\n",
       "      <th>Count</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>2016-01-01</td>\n",
       "      <td>no-event</td>\n",
       "      <td>17</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2016-01-01</td>\n",
       "      <td>no-event</td>\n",
       "      <td>19</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2016-01-01</td>\n",
       "      <td>no-event</td>\n",
       "      <td>20</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>2016-01-02</td>\n",
       "      <td>basketball</td>\n",
       "      <td>16</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>2016-01-02</td>\n",
       "      <td>basketball</td>\n",
       "      <td>18</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       O_date  Event_type  O_hour  Count\n",
       "0  2016-01-01    no-event      17      1\n",
       "1  2016-01-01    no-event      19      1\n",
       "2  2016-01-01    no-event      20      1\n",
       "3  2016-01-02  basketball      16      1\n",
       "4  2016-01-02  basketball      18      1"
      ]
     },
     "execution_count": 40,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "d_test = pd.DataFrame(test.groupby(['O_date','Event_type','O_hour'],as_index=False).size())\n",
    "d_test.reset_index(inplace=True)\n",
    "d_test.columns = ['O_date','Event_type','O_hour','Count']\n",
    "d_test.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "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>date</th>\n",
       "      <th>average temperature</th>\n",
       "      <th>precipitation</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1-1-2016</td>\n",
       "      <td>38.0</td>\n",
       "      <td>0.00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2-1-2016</td>\n",
       "      <td>36.0</td>\n",
       "      <td>0.00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>3-1-2016</td>\n",
       "      <td>40.0</td>\n",
       "      <td>0.00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>4-1-2016</td>\n",
       "      <td>25.0</td>\n",
       "      <td>0.00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>5-1-2016</td>\n",
       "      <td>20.0</td>\n",
       "      <td>0.00</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       date  average temperature precipitation\n",
       "0  1-1-2016                 38.0          0.00\n",
       "1  2-1-2016                 36.0          0.00\n",
       "2  3-1-2016                 40.0          0.00\n",
       "3  4-1-2016                 25.0          0.00\n",
       "4  5-1-2016                 20.0          0.00"
      ]
     },
     "execution_count": 41,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "weather = pd.read_csv('weather_2016.csv')\n",
    "w = weather[['date','average temperature','precipitation']]\n",
    "w.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/Yuwen/anaconda/lib/python3.6/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": [
    "w['date'] = pd.to_datetime(w.date)\n",
    "d_train['O_date'] = pd.to_datetime(d_train.O_date)\n",
    "d_test['O_date'] = pd.to_datetime(d_test.O_date)\n",
    "\n",
    "d_train = d_train.merge(w,left_on='O_date',right_on='date',how='left')\n",
    "d_test = d_test.merge(w,left_on='O_date',right_on='date',how='left')\n",
    "d_train.rename(columns={'average temperature':'temperature'},inplace=True)\n",
    "d_test.rename(columns={'average temperature':'temperature'},inplace=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/Yuwen/anaconda/lib/python3.6/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",
      "\n",
      "See the caveats in the documentation: http://pandas.pydata.org/pandas-docs/stable/indexing.html#indexing-view-versus-copy\n",
      "  \n",
      "/Users/Yuwen/anaconda/lib/python3.6/site-packages/ipykernel_launcher.py:5: 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",
      "  \"\"\"\n"
     ]
    }
   ],
   "source": [
    "d_train['precipitation'] = pd.to_numeric(d_train['precipitation'],errors='coerce')\n",
    "d_train['precipitation'][d_train['precipitation'].isnull()] = 0\n",
    "\n",
    "d_test['precipitation'] = pd.to_numeric(d_test['precipitation'],errors='coerce')\n",
    "d_test['precipitation'][d_test['precipitation'].isnull()] = 0"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "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>O_date</th>\n",
       "      <th>Event_type</th>\n",
       "      <th>O_hour</th>\n",
       "      <th>Count</th>\n",
       "      <th>date</th>\n",
       "      <th>temperature</th>\n",
       "      <th>precipitation</th>\n",
       "      <th>weekday</th>\n",
       "      <th>O_month</th>\n",
       "      <th>season</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>2016-01-01</td>\n",
       "      <td>no-event</td>\n",
       "      <td>16</td>\n",
       "      <td>1</td>\n",
       "      <td>2016-01-01</td>\n",
       "      <td>38.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>True</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2016-01-01</td>\n",
       "      <td>no-event</td>\n",
       "      <td>18</td>\n",
       "      <td>1</td>\n",
       "      <td>2016-01-01</td>\n",
       "      <td>38.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>True</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2016-01-01</td>\n",
       "      <td>no-event</td>\n",
       "      <td>21</td>\n",
       "      <td>1</td>\n",
       "      <td>2016-01-01</td>\n",
       "      <td>38.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>True</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>2016-01-01</td>\n",
       "      <td>no-event</td>\n",
       "      <td>22</td>\n",
       "      <td>1</td>\n",
       "      <td>2016-01-01</td>\n",
       "      <td>38.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>True</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>2016-01-01</td>\n",
       "      <td>no-event</td>\n",
       "      <td>23</td>\n",
       "      <td>1</td>\n",
       "      <td>2016-01-01</td>\n",
       "      <td>38.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>True</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "      O_date Event_type  O_hour  Count       date  temperature  precipitation  \\\n",
       "0 2016-01-01   no-event      16      1 2016-01-01         38.0            0.0   \n",
       "1 2016-01-01   no-event      18      1 2016-01-01         38.0            0.0   \n",
       "2 2016-01-01   no-event      21      1 2016-01-01         38.0            0.0   \n",
       "3 2016-01-01   no-event      22      1 2016-01-01         38.0            0.0   \n",
       "4 2016-01-01   no-event      23      1 2016-01-01         38.0            0.0   \n",
       "\n",
       "   weekday  O_month  season  \n",
       "0     True        1       1  \n",
       "1     True        1       1  \n",
       "2     True        1       1  \n",
       "3     True        1       1  \n",
       "4     True        1       1  "
      ]
     },
     "execution_count": 44,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# generate weekday, month, season\n",
    "d_train['weekday'] = [i not in [5,6] for i in d_train['O_date'].dt.weekday.values]\n",
    "d_train['O_month'] = pd.to_datetime(d_train['O_date']).dt.month\n",
    "d_train['season'] = d_train['O_month'].map({1: 1,\n",
    "                              2: 1,\n",
    "                              3: 2,\n",
    "                              4: 2,\n",
    "                              5: 2,\n",
    "                              6: 3,\n",
    "                              7: 3,\n",
    "                              8: 3,\n",
    "                              9: 4,\n",
    "                              10: 4,\n",
    "                              11: 4,\n",
    "                              12: 1})\n",
    "d_train.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "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>O_date</th>\n",
       "      <th>Event_type</th>\n",
       "      <th>O_hour</th>\n",
       "      <th>Count</th>\n",
       "      <th>date</th>\n",
       "      <th>temperature</th>\n",
       "      <th>precipitation</th>\n",
       "      <th>weekday</th>\n",
       "      <th>O_month</th>\n",
       "      <th>season</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>2016-01-01</td>\n",
       "      <td>no-event</td>\n",
       "      <td>17</td>\n",
       "      <td>1</td>\n",
       "      <td>2016-01-01</td>\n",
       "      <td>38.0</td>\n",
       "      <td>0.00</td>\n",
       "      <td>True</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2016-01-01</td>\n",
       "      <td>no-event</td>\n",
       "      <td>19</td>\n",
       "      <td>1</td>\n",
       "      <td>2016-01-01</td>\n",
       "      <td>38.0</td>\n",
       "      <td>0.00</td>\n",
       "      <td>True</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2016-01-01</td>\n",
       "      <td>no-event</td>\n",
       "      <td>20</td>\n",
       "      <td>1</td>\n",
       "      <td>2016-01-01</td>\n",
       "      <td>38.0</td>\n",
       "      <td>0.00</td>\n",
       "      <td>True</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>2016-01-02</td>\n",
       "      <td>basketball</td>\n",
       "      <td>16</td>\n",
       "      <td>1</td>\n",
       "      <td>2016-01-02</td>\n",
       "      <td>51.5</td>\n",
       "      <td>0.01</td>\n",
       "      <td>False</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>2016-01-02</td>\n",
       "      <td>basketball</td>\n",
       "      <td>18</td>\n",
       "      <td>1</td>\n",
       "      <td>2016-01-02</td>\n",
       "      <td>51.5</td>\n",
       "      <td>0.01</td>\n",
       "      <td>False</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "      O_date  Event_type  O_hour  Count       date  temperature  \\\n",
       "0 2016-01-01    no-event      17      1 2016-01-01         38.0   \n",
       "1 2016-01-01    no-event      19      1 2016-01-01         38.0   \n",
       "2 2016-01-01    no-event      20      1 2016-01-01         38.0   \n",
       "3 2016-01-02  basketball      16      1 2016-01-02         51.5   \n",
       "4 2016-01-02  basketball      18      1 2016-01-02         51.5   \n",
       "\n",
       "   precipitation  weekday  O_month  season  \n",
       "0           0.00     True        1       1  \n",
       "1           0.00     True        1       1  \n",
       "2           0.00     True        1       1  \n",
       "3           0.01    False        1       1  \n",
       "4           0.01    False        1       1  "
      ]
     },
     "execution_count": 45,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# generate weekday, month, season\n",
    "d_test['weekday'] = [i not in [5,6] for i in d_test['O_date'].dt.weekday.values]\n",
    "d_test['O_month'] = pd.to_datetime(d_test['O_date']).dt.month\n",
    "d_test['season'] = d_test['O_month'].map({1: 1,\n",
    "                              2: 1,\n",
    "                              3: 2,\n",
    "                              4: 2,\n",
    "                              5: 2,\n",
    "                              6: 3,\n",
    "                              7: 3,\n",
    "                              8: 3,\n",
    "                              9: 4,\n",
    "                              10: 4,\n",
    "                              11: 4,\n",
    "                              12: 1})\n",
    "d_test.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "d_train.to_csv('trainset.csv')\n",
    "d_test.to_csv('testset.csv')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Run regression model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1258\n",
      "1238\n"
     ]
    }
   ],
   "source": [
    "train = train[train['O_hour'] > 18]\n",
    "print(len(train))\n",
    "train = train[train['Event_type'] != 'other']\n",
    "print(len(train))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "537\n",
      "522\n"
     ]
    }
   ],
   "source": [
    "test = test[test['O_hour'] > 18]\n",
    "print(len(test))\n",
    "test = test[test['Event_type'] != 'other']\n",
    "print(len(test))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(1238, 522)"
      ]
     },
     "execution_count": 49,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(train), len(test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<table class=\"simpletable\">\n",
       "<caption>OLS Regression Results</caption>\n",
       "<tr>\n",
       "  <th>Dep. Variable:</th>          <td>Count</td>      <th>  R-squared:         </th> <td>   0.577</td> \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>Model:</th>                   <td>OLS</td>       <th>  Adj. R-squared:    </th> <td>   0.568</td> \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>Method:</th>             <td>Least Squares</td>  <th>  F-statistic:       </th> <td>   66.12</td> \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>Date:</th>             <td>Wed, 13 Dec 2017</td> <th>  Prob (F-statistic):</th> <td>5.83e-206</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>Time:</th>                 <td>13:42:52</td>     <th>  Log-Likelihood:    </th> <td> -4515.2</td> \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>No. Observations:</th>      <td>  1238</td>      <th>  AIC:               </th> <td>   9082.</td> \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>Df Residuals:</th>          <td>  1212</td>      <th>  BIC:               </th> <td>   9216.</td> \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>Df Model:</th>              <td>    25</td>      <th>                     </th>     <td> </td>    \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>Covariance Type:</th>      <td>nonrobust</td>    <th>                     </th>     <td> </td>    \n",
       "</tr>\n",
       "</table>\n",
       "<table class=\"simpletable\">\n",
       "<tr>\n",
       "                                       <td></td>                                         <th>coef</th>     <th>std err</th>      <th>t</th>      <th>P>|t|</th>  <th>[0.025</th>    <th>0.975]</th>  \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>Intercept</th>                                                                  <td>  -30.5540</td> <td>   13.156</td> <td>   -2.322</td> <td> 0.020</td> <td>  -56.366</td> <td>   -4.742</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>weekday[T.True]</th>                                                            <td>    5.0909</td> <td>    0.609</td> <td>    8.362</td> <td> 0.000</td> <td>    3.896</td> <td>    6.285</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>C(Event_type, Treatment(reference='no-event'))[T.basketball]</th>               <td>   33.7526</td> <td>   25.611</td> <td>    1.318</td> <td> 0.188</td> <td>  -16.494</td> <td>   83.999</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>C(Event_type, Treatment(reference='no-event'))[T.boxing]</th>                   <td>  131.0546</td> <td>   67.190</td> <td>    1.950</td> <td> 0.051</td> <td>   -0.768</td> <td>  262.877</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>C(Event_type, Treatment(reference='no-event'))[T.concert]</th>                  <td>  -52.5809</td> <td>   43.461</td> <td>   -1.210</td> <td> 0.227</td> <td> -137.847</td> <td>   32.685</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>C(Event_type, Treatment(reference='no-event'))[T.family]</th>                   <td>   -0.3230</td> <td>    1.422</td> <td>   -0.227</td> <td> 0.820</td> <td>   -3.113</td> <td>    2.467</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>C(Event_type, Treatment(reference='no-event'))[T.hockey]</th>                   <td>   72.2665</td> <td>   37.859</td> <td>    1.909</td> <td> 0.057</td> <td>   -2.010</td> <td>  146.542</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>precipitation</th>                                                              <td>   -2.8840</td> <td>    0.926</td> <td>   -3.113</td> <td> 0.002</td> <td>   -4.701</td> <td>   -1.067</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>temperature</th>                                                                <td>   -0.0074</td> <td>    0.017</td> <td>   -0.430</td> <td> 0.667</td> <td>   -0.041</td> <td>    0.026</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>season</th>                                                                     <td>   52.2398</td> <td>    4.948</td> <td>   10.558</td> <td> 0.000</td> <td>   42.532</td> <td>   61.947</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>season:C(Event_type, Treatment(reference='no-event'))[T.basketball]</th>        <td>  -19.4330</td> <td>    9.633</td> <td>   -2.017</td> <td> 0.044</td> <td>  -38.333</td> <td>   -0.533</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>season:C(Event_type, Treatment(reference='no-event'))[T.boxing]</th>            <td>  -70.2241</td> <td>   28.438</td> <td>   -2.469</td> <td> 0.014</td> <td> -126.018</td> <td>  -14.430</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>season:C(Event_type, Treatment(reference='no-event'))[T.concert]</th>           <td>   14.6598</td> <td>   15.156</td> <td>    0.967</td> <td> 0.334</td> <td>  -15.076</td> <td>   44.396</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>season:C(Event_type, Treatment(reference='no-event'))[T.family]</th>            <td>   -1.5080</td> <td>    5.701</td> <td>   -0.265</td> <td> 0.791</td> <td>  -12.693</td> <td>    9.677</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>season:C(Event_type, Treatment(reference='no-event'))[T.hockey]</th>            <td>  -11.5739</td> <td>   12.001</td> <td>   -0.964</td> <td> 0.335</td> <td>  -35.120</td> <td>   11.972</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>O_hour</th>                                                                     <td>    1.2097</td> <td>    0.626</td> <td>    1.932</td> <td> 0.054</td> <td>   -0.019</td> <td>    2.438</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>O_hour:C(Event_type, Treatment(reference='no-event'))[T.basketball]</th>        <td>   -1.3756</td> <td>    1.227</td> <td>   -1.121</td> <td> 0.262</td> <td>   -3.782</td> <td>    1.031</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>O_hour:C(Event_type, Treatment(reference='no-event'))[T.boxing]</th>            <td>   -5.8264</td> <td>    3.215</td> <td>   -1.813</td> <td> 0.070</td> <td>  -12.133</td> <td>    0.480</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>O_hour:C(Event_type, Treatment(reference='no-event'))[T.concert]</th>           <td>    2.1891</td> <td>    2.076</td> <td>    1.054</td> <td> 0.292</td> <td>   -1.884</td> <td>    6.262</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>O_hour:C(Event_type, Treatment(reference='no-event'))[T.family]</th>            <td>    0.9904</td> <td>    0.674</td> <td>    1.470</td> <td> 0.142</td> <td>   -0.332</td> <td>    2.312</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>O_hour:C(Event_type, Treatment(reference='no-event'))[T.hockey]</th>            <td>   -2.8423</td> <td>    1.817</td> <td>   -1.564</td> <td> 0.118</td> <td>   -6.407</td> <td>    0.722</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>season:O_hour</th>                                                              <td>   -2.1335</td> <td>    0.236</td> <td>   -9.042</td> <td> 0.000</td> <td>   -2.596</td> <td>   -1.671</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>season:O_hour:C(Event_type, Treatment(reference='no-event'))[T.basketball]</th> <td>    0.8599</td> <td>    0.459</td> <td>    1.873</td> <td> 0.061</td> <td>   -0.041</td> <td>    1.761</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>season:O_hour:C(Event_type, Treatment(reference='no-event'))[T.boxing]</th>     <td>    3.1822</td> <td>    1.350</td> <td>    2.357</td> <td> 0.019</td> <td>    0.533</td> <td>    5.831</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>season:O_hour:C(Event_type, Treatment(reference='no-event'))[T.concert]</th>    <td>   -0.5890</td> <td>    0.725</td> <td>   -0.813</td> <td> 0.417</td> <td>   -2.011</td> <td>    0.833</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>season:O_hour:C(Event_type, Treatment(reference='no-event'))[T.family]</th>     <td>   -0.1418</td> <td>    0.304</td> <td>   -0.466</td> <td> 0.641</td> <td>   -0.738</td> <td>    0.455</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>season:O_hour:C(Event_type, Treatment(reference='no-event'))[T.hockey]</th>     <td>    0.4689</td> <td>    0.573</td> <td>    0.818</td> <td> 0.413</td> <td>   -0.655</td> <td>    1.593</td>\n",
       "</tr>\n",
       "</table>\n",
       "<table class=\"simpletable\">\n",
       "<tr>\n",
       "  <th>Omnibus:</th>       <td>171.471</td> <th>  Durbin-Watson:     </th> <td>   2.020</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>Prob(Omnibus):</th> <td> 0.000</td>  <th>  Jarque-Bera (JB):  </th> <td> 373.233</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>Skew:</th>          <td> 0.802</td>  <th>  Prob(JB):          </th> <td>8.99e-82</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>Kurtosis:</th>      <td> 5.159</td>  <th>  Cond. No.          </th> <td>1.04e+16</td>\n",
       "</tr>\n",
       "</table>"
      ],
      "text/plain": [
       "<class 'statsmodels.iolib.summary.Summary'>\n",
       "\"\"\"\n",
       "                            OLS Regression Results                            \n",
       "==============================================================================\n",
       "Dep. Variable:                  Count   R-squared:                       0.577\n",
       "Model:                            OLS   Adj. R-squared:                  0.568\n",
       "Method:                 Least Squares   F-statistic:                     66.12\n",
       "Date:                Wed, 13 Dec 2017   Prob (F-statistic):          5.83e-206\n",
       "Time:                        13:42:52   Log-Likelihood:                -4515.2\n",
       "No. Observations:                1238   AIC:                             9082.\n",
       "Df Residuals:                    1212   BIC:                             9216.\n",
       "Df Model:                          25                                         \n",
       "Covariance Type:            nonrobust                                         \n",
       "==============================================================================================================================================\n",
       "                                                                                 coef    std err          t      P>|t|      [0.025      0.975]\n",
       "----------------------------------------------------------------------------------------------------------------------------------------------\n",
       "Intercept                                                                    -30.5540     13.156     -2.322      0.020     -56.366      -4.742\n",
       "weekday[T.True]                                                                5.0909      0.609      8.362      0.000       3.896       6.285\n",
       "C(Event_type, Treatment(reference='no-event'))[T.basketball]                  33.7526     25.611      1.318      0.188     -16.494      83.999\n",
       "C(Event_type, Treatment(reference='no-event'))[T.boxing]                     131.0546     67.190      1.950      0.051      -0.768     262.877\n",
       "C(Event_type, Treatment(reference='no-event'))[T.concert]                    -52.5809     43.461     -1.210      0.227    -137.847      32.685\n",
       "C(Event_type, Treatment(reference='no-event'))[T.family]                      -0.3230      1.422     -0.227      0.820      -3.113       2.467\n",
       "C(Event_type, Treatment(reference='no-event'))[T.hockey]                      72.2665     37.859      1.909      0.057      -2.010     146.542\n",
       "precipitation                                                                 -2.8840      0.926     -3.113      0.002      -4.701      -1.067\n",
       "temperature                                                                   -0.0074      0.017     -0.430      0.667      -0.041       0.026\n",
       "season                                                                        52.2398      4.948     10.558      0.000      42.532      61.947\n",
       "season:C(Event_type, Treatment(reference='no-event'))[T.basketball]          -19.4330      9.633     -2.017      0.044     -38.333      -0.533\n",
       "season:C(Event_type, Treatment(reference='no-event'))[T.boxing]              -70.2241     28.438     -2.469      0.014    -126.018     -14.430\n",
       "season:C(Event_type, Treatment(reference='no-event'))[T.concert]              14.6598     15.156      0.967      0.334     -15.076      44.396\n",
       "season:C(Event_type, Treatment(reference='no-event'))[T.family]               -1.5080      5.701     -0.265      0.791     -12.693       9.677\n",
       "season:C(Event_type, Treatment(reference='no-event'))[T.hockey]              -11.5739     12.001     -0.964      0.335     -35.120      11.972\n",
       "O_hour                                                                         1.2097      0.626      1.932      0.054      -0.019       2.438\n",
       "O_hour:C(Event_type, Treatment(reference='no-event'))[T.basketball]           -1.3756      1.227     -1.121      0.262      -3.782       1.031\n",
       "O_hour:C(Event_type, Treatment(reference='no-event'))[T.boxing]               -5.8264      3.215     -1.813      0.070     -12.133       0.480\n",
       "O_hour:C(Event_type, Treatment(reference='no-event'))[T.concert]               2.1891      2.076      1.054      0.292      -1.884       6.262\n",
       "O_hour:C(Event_type, Treatment(reference='no-event'))[T.family]                0.9904      0.674      1.470      0.142      -0.332       2.312\n",
       "O_hour:C(Event_type, Treatment(reference='no-event'))[T.hockey]               -2.8423      1.817     -1.564      0.118      -6.407       0.722\n",
       "season:O_hour                                                                 -2.1335      0.236     -9.042      0.000      -2.596      -1.671\n",
       "season:O_hour:C(Event_type, Treatment(reference='no-event'))[T.basketball]     0.8599      0.459      1.873      0.061      -0.041       1.761\n",
       "season:O_hour:C(Event_type, Treatment(reference='no-event'))[T.boxing]         3.1822      1.350      2.357      0.019       0.533       5.831\n",
       "season:O_hour:C(Event_type, Treatment(reference='no-event'))[T.concert]       -0.5890      0.725     -0.813      0.417      -2.011       0.833\n",
       "season:O_hour:C(Event_type, Treatment(reference='no-event'))[T.family]        -0.1418      0.304     -0.466      0.641      -0.738       0.455\n",
       "season:O_hour:C(Event_type, Treatment(reference='no-event'))[T.hockey]         0.4689      0.573      0.818      0.413      -0.655       1.593\n",
       "==============================================================================\n",
       "Omnibus:                      171.471   Durbin-Watson:                   2.020\n",
       "Prob(Omnibus):                  0.000   Jarque-Bera (JB):              373.233\n",
       "Skew:                           0.802   Prob(JB):                     8.99e-82\n",
       "Kurtosis:                       5.159   Cond. No.                     1.04e+16\n",
       "==============================================================================\n",
       "\n",
       "Warnings:\n",
       "[1] Standard Errors assume that the covariance matrix of the errors is correctly specified.\n",
       "[2] The smallest eigenvalue is 8.24e-26. This might indicate that there are\n",
       "strong multicollinearity problems or that the design matrix is singular.\n",
       "\"\"\""
      ]
     },
     "execution_count": 50,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "lm_train = smf.ols(formula=\"Count ~ weekday + precipitation + temperature + season * O_hour * C(Event_type, Treatment(reference='no-event'))\",data=train).fit()\n",
    "lm_train.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Index(['Unnamed: 0', 'O_date', 'Event_type', 'O_hour', 'Count', 'date',\n",
       "       'temperature', 'precipitation', 'weekday', 'O_month', 'season'],\n",
       "      dtype='object')"
      ]
     },
     "execution_count": 51,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "test.keys()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "metadata": {},
   "outputs": [],
   "source": [
    "pre = lm_train.predict(test[[\"weekday\",\"precipitation\",\"temperature\",\"season\",\"O_hour\",\"Event_type\"]])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[<matplotlib.lines.Line2D at 0x10f2a87b8>]"
      ]
     },
     "execution_count": 54,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAABIEAAAHVCAYAAABmEeuHAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x10f2b8e10>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "x = range(len(test))\n",
    "pl.figure(figsize=(20,8))\n",
    "pl.plot(x,pre,'.')\n",
    "pl.plot(x,test.Count,'.')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# For displaying R^2 and linear model summary\n",
    "def modelEval(lm, key = 'Count'):\n",
    "    lmy = lm.predict(test[[\"weekday\",\"precipitation\",\"temperature\",\"season\",\"O_hour\",\"Event_type\"]])\n",
    "    y_err = (lmy - test[key]).reshape(test[key].shape[0],)\n",
    "    y_norm = (test[key]-np.mean(test[key])).reshape(test[key].shape[0],)\n",
    "    \n",
    "    #y_err = (predicted - test_y).reshape(test_y.shape[0],)\n",
    "    #y_norm = (test_y-np.mean(test_y)).reshape(test_y.shape[0],)\n",
    "    #r2 = 1-y_err.dot(y_err)/y_norm.dot(y_norm)\n",
    "    return 1-y_err.dot(y_err)/y_norm.dot(y_norm)  # R2 w.r.t. test set"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "In sample R-squared: 0.576950\n",
      "Out of sample R-squared: 0.611713\n",
      "Fitted coefficients:\n",
      "Intercept                                                                     -30.553982\n",
      "weekday[T.True]                                                                 5.090870\n",
      "C(Event_type, Treatment(reference='no-event'))[T.basketball]                   33.752583\n",
      "C(Event_type, Treatment(reference='no-event'))[T.boxing]                      131.054608\n",
      "C(Event_type, Treatment(reference='no-event'))[T.concert]                     -52.580947\n",
      "C(Event_type, Treatment(reference='no-event'))[T.family]                       -0.323011\n",
      "C(Event_type, Treatment(reference='no-event'))[T.hockey]                       72.266471\n",
      "precipitation                                                                  -2.883957\n",
      "temperature                                                                    -0.007381\n",
      "season                                                                         52.239762\n",
      "season:C(Event_type, Treatment(reference='no-event'))[T.basketball]           -19.432997\n",
      "season:C(Event_type, Treatment(reference='no-event'))[T.boxing]               -70.224134\n",
      "season:C(Event_type, Treatment(reference='no-event'))[T.concert]               14.659777\n",
      "season:C(Event_type, Treatment(reference='no-event'))[T.family]                -1.508004\n",
      "season:C(Event_type, Treatment(reference='no-event'))[T.hockey]               -11.573887\n",
      "O_hour                                                                          1.209673\n",
      "O_hour:C(Event_type, Treatment(reference='no-event'))[T.basketball]            -1.375557\n",
      "O_hour:C(Event_type, Treatment(reference='no-event'))[T.boxing]                -5.826358\n",
      "O_hour:C(Event_type, Treatment(reference='no-event'))[T.concert]                2.189094\n",
      "O_hour:C(Event_type, Treatment(reference='no-event'))[T.family]                 0.990368\n",
      "O_hour:C(Event_type, Treatment(reference='no-event'))[T.hockey]                -2.842295\n",
      "season:O_hour                                                                  -2.133492\n",
      "season:O_hour:C(Event_type, Treatment(reference='no-event'))[T.basketball]      0.859925\n",
      "season:O_hour:C(Event_type, Treatment(reference='no-event'))[T.boxing]          3.182158\n",
      "season:O_hour:C(Event_type, Treatment(reference='no-event'))[T.concert]        -0.588962\n",
      "season:O_hour:C(Event_type, Treatment(reference='no-event'))[T.family]         -0.141758\n",
      "season:O_hour:C(Event_type, Treatment(reference='no-event'))[T.hockey]          0.468946\n",
      "dtype: float64\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/Yuwen/anaconda/lib/python3.6/site-packages/ipykernel_launcher.py:4: FutureWarning: reshape is deprecated and will raise in a subsequent release. Please use .values.reshape(...) instead\n",
      "  after removing the cwd from sys.path.\n",
      "/Users/Yuwen/anaconda/lib/python3.6/site-packages/ipykernel_launcher.py:5: FutureWarning: reshape is deprecated and will raise in a subsequent release. Please use .values.reshape(...) instead\n",
      "  \"\"\"\n"
     ]
    }
   ],
   "source": [
    "#result = smf.ols(formula = 'Y ~ '+ '+'.join(data3.columns[:-1]), data = train).fit()\n",
    "print('In sample R-squared: %f'%lm_train.rsquared)\n",
    "print('Out of sample R-squared: %f'%modelEval(lm_train))\n",
    "print('Fitted coefficients:')\n",
    "print(lm_train.params)"
   ]
  },
  {
   "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
}