swiftly mnl v1.3.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "import sys, os, copy, itertools\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "import statsmodels.api as st\n",
    "import statsmodels.formula.api as smf\n",
    "import matplotlib.pyplot as plt"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "INDIR = r'Q:\\Model Research\\Swiftly\\processed\\v0.6'\n",
    "INFILE = r'choices_with_valid_rideshare.csv'\n",
    "choices = pd.read_csv(os.path.join(INDIR,INFILE))\n",
    "# drop duplicate listings within a search_id\n",
    "choices = choices.drop_duplicates(subset=['search_id','travel_provider','travel_product'])\n",
    "# filter to just records where a choice is made\n",
    "chosen_search_ids = choices.loc[choices['chosen']==1,'search_id'].tolist()\n",
    "choices = choices[choices['search_id'].isin(chosen_search_ids)]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "# clean data using 'drop' flag\n",
    "choices['drop'] = 0\n",
    "# flag records with invalid walk time or travel time\n",
    "choices.loc[(choices['travel_mode'] == 'transit') & ((choices['travel_initial_walk_time'] == -1) | (pd.isnull(choices['travel_initial_walk_time']))),'drop'] = 1\n",
    "choices.loc[choices['travel_time'] == 0,'drop'] = 1\n",
    "# flag records with both trip-ends outside of 9 counties\n",
    "choices.loc[(choices['county_start']=='Other') & (choices['county_end']=='Other'),'drop'] = 1\n",
    "# flag records with invalid cost\n",
    "choices.loc[(choices['travel_mode'] == 'transit') & (choices['travel_price']==-1),'drop'] = 1\n",
    "# drop flagged records\n",
    "choices = choices[choices['drop']==0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# get walk and wait time in minutes to next 3 arrivals\n",
    "choices['search_time'] = pd.to_datetime(choices['timestamp'])\n",
    "choices['arr1'] = pd.to_datetime(choices['travel_arrival_time_1'])\n",
    "choices['arr2'] = pd.to_datetime(choices['travel_arrival_time_2'])\n",
    "choices['arr3'] = pd.to_datetime(choices['travel_arrival_time_3'])\n",
    "choices['wait1'] = choices['arr1'] - choices['search_time']\n",
    "choices['wait2'] = choices['arr2'] - choices['search_time']\n",
    "choices['wait3'] = choices['arr3'] - choices['search_time']\n",
    "choices['wait1'] = choices['wait1'].map(lambda x:x.total_seconds() / 60.0)\n",
    "choices['wait2'] = choices['wait2'].map(lambda x:x.total_seconds() / 60.0)\n",
    "choices['wait3'] = choices['wait3'].map(lambda x:x.total_seconds() / 60.0)\n",
    "choices['walk_time'] = choices['travel_initial_walk_time'] / 60.0\n",
    "# calc headways in minutes\n",
    "choices['h1_2'] = choices['wait2'] - choices['wait1']\n",
    "choices['h2_3'] = choices['wait3'] - choices['wait2']\n",
    "choices['h3_3'] = choices['wait3'] - choices['wait1']\n",
    "# pick first wait time that the user has time to walk to\n",
    "choices['wait'] = choices['wait1']\n",
    "choices.loc[(choices['wait'] < choices['walk_time']) & (pd.notnull(choices['wait2'])),'wait'] = choices['wait2']\n",
    "choices.loc[(choices['wait'] < choices['walk_time']) & (pd.notnull(choices['wait3'])),'wait'] = choices['wait3']\n",
    "choices['wait'] = choices['wait'] - choices['walk_time']\n",
    "choices['headway'] = 99\n",
    "choices.loc[pd.notnull(choices['h1_2']),'headway'] = choices['h1_2']\n",
    "choices['n_arrivals'] = 0\n",
    "choices.loc[pd.notnull(choices['arr1']),'n_arrivals'] += 1\n",
    "choices.loc[pd.notnull(choices['arr2']),'n_arrivals'] += 1\n",
    "choices.loc[pd.notnull(choices['arr3']),'n_arrivals'] += 1\n",
    "# rename pool; it shows up as pool and uberpool in the data\n",
    "choices.loc[choices['travel_product']=='pool','travel_product'] = 'uberpool'\n",
    "# classify uber products into 4 types\n",
    "choices['uber_mode_class'] = np.nan\n",
    "choices.loc[choices['travel_product'].isin(['uberx','uberxl']),'uber_mode_class'] = 'basic'\n",
    "choices.loc[choices['travel_product'].isin(['uberselect','uberblack','ubersuv']),'uber_mode_class'] = 'premium'\n",
    "choices.loc[choices['travel_product']=='uberpool','uber_mode_class'] = 'pool'\n",
    "choices.loc[choices['travel_product']=='ubertaxi','uber_mode_class'] = 'taxi'\n",
    "choices.loc[(pd.isnull(choices['uber_mode_class'])) & (choices['travel_provider']=='Uber'),'uber_mode_class'] = 'other'\n",
    "# classify uber products into 2 types\n",
    "choices['uber_mode_class2'] = np.nan\n",
    "choices.loc[choices['travel_product'].isin(['uberx','uberxl','pool','taxi']),'uber_mode_class2'] = 'basic'\n",
    "choices.loc[choices['travel_product'].isin(['uberselect','uberblack','ubersuv']),'uber_mode_class2'] = 'premium'\n",
    "choices.loc[(pd.isnull(choices['uber_mode_class'])) & (choices['travel_provider']=='Uber'),'uber_mode_class2'] = 'basic'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Python27_64\\lib\\site-packages\\ipython-5.0.0b3-py2.7.egg\\IPython\\core\\interactiveshell.py:2765: DtypeWarning: Columns (0,46,47) have mixed types. Specify dtype option on import or set low_memory=False.\n",
      "  interactivity=interactivity, compiler=compiler, result=result)\n"
     ]
    }
   ],
   "source": [
    "user_info = pd.read_csv(r'Q:\\Model Research\\Swiftly\\processed\\v0.6\\user_data\\users_with_demographics.csv')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "user_info = user_info.set_index('user_id')\n",
    "choices = choices.set_index('user_id')\n",
    "choices['master_user_id'] = user_info['master_user_id']\n",
    "choices['gender'] = user_info['gender']\n",
    "choices['age'] = user_info['age']\n",
    "choices['age_cat'] = user_info['age_cat']\n",
    "choices['hhinc_cat'] = user_info['hhinc_cat']\n",
    "choices['race'] = user_info['race']\n",
    "choices['employ'] = user_info['employ']\n",
    "choices['hhinc'] = user_info['hhinc']\n",
    "choices['hhsize'] = user_info['hhsize']\n",
    "choices['hhvehs'] = user_info['hhvehs']\n",
    "choices['female'] = 0\n",
    "choices.loc[choices['gender']==2,'female'] = 1\n",
    "choices = choices.reset_index()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "collapsed": false,
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "chosen_mode3\n",
      "biking       107\n",
      "scoot        348\n",
      "transit    19189\n",
      "uber         779\n",
      "walking     1571\n",
      "dtype: int64\n"
     ]
    }
   ],
   "source": [
    "# set up estimation set.\n",
    "estimation_set = pd.DataFrame(index=choices['search_id'].drop_duplicates())\n",
    "chosen_mode = choices[choices['chosen']==1]\n",
    "chosen_mode = chosen_mode.set_index('search_id')\n",
    "estimation_set['chosen_mode'] = chosen_mode['travel_mode']\n",
    "estimation_set['chosen_mode2'] = chosen_mode['travel_mode']\n",
    "estimation_set['chosen_mode3'] = chosen_mode['travel_mode']\n",
    "estimation_set.loc[estimation_set['chosen_mode']=='rideshare','chosen_mode'] = chosen_mode['uber_mode_class']\n",
    "estimation_set.loc[estimation_set['chosen_mode2']=='rideshare','chosen_mode2'] = chosen_mode['uber_mode_class2']\n",
    "estimation_set.loc[estimation_set['chosen_mode3']=='rideshare','chosen_mode3'] = \"uber\"\n",
    "\n",
    "estimation_set = estimation_set.dropna()\n",
    "print estimation_set.groupby(['chosen_mode3']).count().max(axis=1)\n",
    "\n",
    "estimation_set['start_is_near_home'] = chosen_mode['start_is_near_home'].replace(-1, np.nan)\n",
    "estimation_set['start_is_near_work'] = chosen_mode['start_is_near_work'].replace(-1, np.nan)\n",
    "estimation_set['end_is_near_home'] = chosen_mode['end_is_near_home'].replace(-1, np.nan)\n",
    "estimation_set['end_is_near_work'] = chosen_mode['end_is_near_work'].replace(-1, np.nan)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "collapsed": false,
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "['age_cat_18-24', 'age_cat_25-34', 'age_cat_35-44', 'age_cat_45-54', 'age_cat_55-64', 'age_cat_65+', 'age_cat_nan']\n",
      "['hhinc_cat_$25-$50', 'hhinc_cat_$50-$75', 'hhinc_cat_$75+', 'hhinc_cat_nan']\n",
      "['employ_2.0', 'employ_3.0', 'employ_4.0', 'employ_5.0', 'employ_nan']\n"
     ]
    }
   ],
   "source": [
    "# add demographcis to estimation set\n",
    "estimation_set['master_user_id'] = chosen_mode['master_user_id']\n",
    "estimation_set['age_cat'] = chosen_mode['age_cat']\n",
    "estimation_set['age'] = chosen_mode['age']\n",
    "estimation_set['hhinc_cat'] = chosen_mode['hhinc_cat']\n",
    "estimation_set['hhinc'] = chosen_mode['hhinc']\n",
    "estimation_set['employ'] = chosen_mode['employ']\n",
    "estimation_set['hhvehs'] = chosen_mode['hhvehs']\n",
    "estimation_set['female'] = chosen_mode['female']\n",
    "\n",
    "# convert to dummies\n",
    "estimation_set = pd.get_dummies(estimation_set,columns=['master_user_id','age_cat','hhinc_cat','employ'], dummy_na=True, drop_first=True)\n",
    "\n",
    "# get dummy columns\n",
    "user_dummy_cols = []\n",
    "age_dummy_cols = []\n",
    "inc_dummy_cols = []\n",
    "employ_dummy_cols = []\n",
    "for col in estimation_set.columns.tolist():\n",
    "    if col[:14] == 'master_user_id':\n",
    "        user_dummy_cols.append(col)\n",
    "    elif col[:7] == 'age_cat':\n",
    "        age_dummy_cols.append(col)\n",
    "    elif col[:9] == 'hhinc_cat':\n",
    "        inc_dummy_cols.append(col)\n",
    "    elif col[:6] == 'employ':\n",
    "        employ_dummy_cols.append(col)\n",
    "        \n",
    "estimation_set['master_user_id'] = chosen_mode['master_user_id']\n",
    "print age_dummy_cols\n",
    "print inc_dummy_cols\n",
    "print employ_dummy_cols"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "travel_provider\n",
       "Alcatraz Hornblower Ferry                            0.0\n",
       "Altamont Commuter Express                            0.0\n",
       "Bay Area Rapid Transit                            1917.0\n",
       "Bear Transit - UC Berkeley Shuttle                   6.0\n",
       "Blue & Gold Fleet                                    1.0\n",
       "Caltrain                                           109.0\n",
       "Commute.org Shuttle                                  6.0\n",
       "County Connection                                   13.0\n",
       "Golden Gate Ferry                                    0.0\n",
       "Marin Transit                                        6.0\n",
       "Mountain View Community Shuttle                      3.0\n",
       "PresidiGo Shuttle                                   37.0\n",
       "SamTrans                                           151.0\n",
       "San Francisco Bay Ferry                              2.0\n",
       "San Francisco Municipal Transportation Agency    16428.0\n",
       "Santa Cruz Metro                                     5.0\n",
       "SolTrans                                             2.0\n",
       "VTA                                                500.0\n",
       "WestCat (Western Contra Costa)                       3.0\n",
       "Name: chosen, dtype: float64"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "transit_options = choices[choices['travel_mode']=='transit']\n",
    "transit_options.groupby(['travel_provider'])['chosen'].sum()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "collapsed": false,
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "trn_sch_group = transit_options.groupby(['search_id'])\n",
    "best_transit = transit_options.loc[trn_sch_group['travel_time'].idxmin(),:]\n",
    "best_transit = best_transit.set_index('search_id')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "collapsed": false,
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "estimation_set['transit_cost'] = best_transit['travel_price']\n",
    "estimation_set['transit_time'] = best_transit['travel_time'] / 60.0\n",
    "estimation_set['transit_dist'] = best_transit['travel_distance'] / 5280.0\n",
    "estimation_set['wait'] = best_transit['wait']\n",
    "estimation_set['headway'] = best_transit['headway']\n",
    "estimation_set['walk'] = best_transit['walk_time']\n",
    "estimation_set['n_arr'] = best_transit['n_arrivals']\n",
    "n_transit = transit_options.groupby(['search_id']).count().max(axis=1)\n",
    "estimation_set['n_transit'] = n_transit\n",
    "estimation_set = estimation_set.dropna(subset=['transit_cost','transit_time','transit_dist'])\n",
    "estimation_set['wait'] = estimation_set['wait'].fillna(999)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "collapsed": false,
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "uber_mode_class\n",
       "basic      444.0\n",
       "other       43.0\n",
       "pool       363.0\n",
       "premium    335.0\n",
       "taxi        49.0\n",
       "Name: chosen, dtype: float64"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "choices.groupby('uber_mode_class')['chosen'].sum()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "collapsed": false,
    "scrolled": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "travel_provider  travel_product\n",
       "Uber             assist             37.0\n",
       "                 uberblack         154.0\n",
       "                 uberpool          363.0\n",
       "                 uberselect        103.0\n",
       "                 ubersuv            78.0\n",
       "                 ubertaxi           49.0\n",
       "                 uberwav             6.0\n",
       "                 uberx             373.0\n",
       "                 uberxl             71.0\n",
       "Name: chosen, dtype: float64"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "g = choices.groupby(['travel_provider','travel_product'])\n",
    "s = g['chosen'].sum()\n",
    "s.loc[['Uber',]]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "uber_grouped = choices[choices['travel_provider']=='Uber'].groupby(['search_id','uber_mode_class'])\n",
    "best_uber = uber_grouped['travel_price'].idxmin()\n",
    "best_uber_idx = copy.deepcopy(best_uber)\n",
    "best_uber = choices.loc[best_uber_idx,['search_id','travel_provider','uber_mode_class','travel_product','travel_price','travel_distance','travel_time']]\n",
    "price = best_uber.pivot(index='search_id',columns='uber_mode_class',values='travel_price')\n",
    "dist = best_uber.pivot(index='search_id',columns='uber_mode_class',values='travel_distance')\n",
    "time = best_uber.pivot(index='search_id',columns='uber_mode_class',values='travel_time')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "collapsed": false,
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>search_id</th>\n",
       "      <th>travel_provider</th>\n",
       "      <th>uber_mode_class</th>\n",
       "      <th>travel_product</th>\n",
       "      <th>travel_price</th>\n",
       "      <th>travel_distance</th>\n",
       "      <th>travel_time</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>224920</th>\n",
       "      <td>309474</td>\n",
       "      <td>Uber</td>\n",
       "      <td>basic</td>\n",
       "      <td>uberx</td>\n",
       "      <td>6.55</td>\n",
       "      <td>2144.0</td>\n",
       "      <td>393.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>224921</th>\n",
       "      <td>309474</td>\n",
       "      <td>Uber</td>\n",
       "      <td>premium</td>\n",
       "      <td>uberselect</td>\n",
       "      <td>13.49</td>\n",
       "      <td>2144.0</td>\n",
       "      <td>393.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>224925</th>\n",
       "      <td>309474</td>\n",
       "      <td>Uber</td>\n",
       "      <td>taxi</td>\n",
       "      <td>ubertaxi</td>\n",
       "      <td>6.55</td>\n",
       "      <td>2144.0</td>\n",
       "      <td>393.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>100855</th>\n",
       "      <td>309475</td>\n",
       "      <td>Uber</td>\n",
       "      <td>basic</td>\n",
       "      <td>uberx</td>\n",
       "      <td>7.48</td>\n",
       "      <td>3335.0</td>\n",
       "      <td>422.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>100856</th>\n",
       "      <td>309475</td>\n",
       "      <td>Uber</td>\n",
       "      <td>premium</td>\n",
       "      <td>uberselect</td>\n",
       "      <td>15.77</td>\n",
       "      <td>3335.0</td>\n",
       "      <td>422.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>78546</th>\n",
       "      <td>309477</td>\n",
       "      <td>Uber</td>\n",
       "      <td>basic</td>\n",
       "      <td>uberx</td>\n",
       "      <td>38.95</td>\n",
       "      <td>39564.0</td>\n",
       "      <td>1945.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>78547</th>\n",
       "      <td>309477</td>\n",
       "      <td>Uber</td>\n",
       "      <td>premium</td>\n",
       "      <td>uberselect</td>\n",
       "      <td>90.36</td>\n",
       "      <td>39564.0</td>\n",
       "      <td>1945.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>59904</th>\n",
       "      <td>309481</td>\n",
       "      <td>Uber</td>\n",
       "      <td>basic</td>\n",
       "      <td>uberx</td>\n",
       "      <td>22.74</td>\n",
       "      <td>20872.0</td>\n",
       "      <td>1166.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>59907</th>\n",
       "      <td>309481</td>\n",
       "      <td>Uber</td>\n",
       "      <td>premium</td>\n",
       "      <td>uberselect</td>\n",
       "      <td>51.93</td>\n",
       "      <td>20872.0</td>\n",
       "      <td>1166.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>310828</th>\n",
       "      <td>309487</td>\n",
       "      <td>Uber</td>\n",
       "      <td>basic</td>\n",
       "      <td>uberx</td>\n",
       "      <td>15.29</td>\n",
       "      <td>10647.0</td>\n",
       "      <td>1128.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>310830</th>\n",
       "      <td>309487</td>\n",
       "      <td>Uber</td>\n",
       "      <td>premium</td>\n",
       "      <td>uberselect</td>\n",
       "      <td>34.14</td>\n",
       "      <td>10647.0</td>\n",
       "      <td>1128.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>310834</th>\n",
       "      <td>309487</td>\n",
       "      <td>Uber</td>\n",
       "      <td>taxi</td>\n",
       "      <td>ubertaxi</td>\n",
       "      <td>15.29</td>\n",
       "      <td>10647.0</td>\n",
       "      <td>1128.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>182706</th>\n",
       "      <td>309488</td>\n",
       "      <td>Uber</td>\n",
       "      <td>basic</td>\n",
       "      <td>uberx</td>\n",
       "      <td>8.63</td>\n",
       "      <td>3521.0</td>\n",
       "      <td>699.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>182708</th>\n",
       "      <td>309488</td>\n",
       "      <td>Uber</td>\n",
       "      <td>premium</td>\n",
       "      <td>uberselect</td>\n",
       "      <td>18.39</td>\n",
       "      <td>3521.0</td>\n",
       "      <td>699.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>182704</th>\n",
       "      <td>309488</td>\n",
       "      <td>Uber</td>\n",
       "      <td>taxi</td>\n",
       "      <td>ubertaxi</td>\n",
       "      <td>8.63</td>\n",
       "      <td>3521.0</td>\n",
       "      <td>699.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>310837</th>\n",
       "      <td>309490</td>\n",
       "      <td>Uber</td>\n",
       "      <td>basic</td>\n",
       "      <td>uberx</td>\n",
       "      <td>15.30</td>\n",
       "      <td>10636.0</td>\n",
       "      <td>1132.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>310840</th>\n",
       "      <td>309490</td>\n",
       "      <td>Uber</td>\n",
       "      <td>premium</td>\n",
       "      <td>uberselect</td>\n",
       "      <td>34.16</td>\n",
       "      <td>10636.0</td>\n",
       "      <td>1132.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>164320</th>\n",
       "      <td>309505</td>\n",
       "      <td>Uber</td>\n",
       "      <td>basic</td>\n",
       "      <td>uberx</td>\n",
       "      <td>8.07</td>\n",
       "      <td>3859.0</td>\n",
       "      <td>481.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>164323</th>\n",
       "      <td>309505</td>\n",
       "      <td>Uber</td>\n",
       "      <td>premium</td>\n",
       "      <td>uberselect</td>\n",
       "      <td>17.15</td>\n",
       "      <td>3859.0</td>\n",
       "      <td>481.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>264696</th>\n",
       "      <td>309512</td>\n",
       "      <td>Uber</td>\n",
       "      <td>basic</td>\n",
       "      <td>uberx</td>\n",
       "      <td>12.17</td>\n",
       "      <td>7008.0</td>\n",
       "      <td>984.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>264699</th>\n",
       "      <td>309512</td>\n",
       "      <td>Uber</td>\n",
       "      <td>premium</td>\n",
       "      <td>uberselect</td>\n",
       "      <td>26.73</td>\n",
       "      <td>7008.0</td>\n",
       "      <td>984.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>264692</th>\n",
       "      <td>309512</td>\n",
       "      <td>Uber</td>\n",
       "      <td>taxi</td>\n",
       "      <td>ubertaxi</td>\n",
       "      <td>12.17</td>\n",
       "      <td>7008.0</td>\n",
       "      <td>984.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>273201</th>\n",
       "      <td>309516</td>\n",
       "      <td>Uber</td>\n",
       "      <td>basic</td>\n",
       "      <td>uberx</td>\n",
       "      <td>6.55</td>\n",
       "      <td>1967.0</td>\n",
       "      <td>320.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>273197</th>\n",
       "      <td>309516</td>\n",
       "      <td>Uber</td>\n",
       "      <td>premium</td>\n",
       "      <td>uberselect</td>\n",
       "      <td>12.58</td>\n",
       "      <td>1967.0</td>\n",
       "      <td>320.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>273202</th>\n",
       "      <td>309516</td>\n",
       "      <td>Uber</td>\n",
       "      <td>taxi</td>\n",
       "      <td>ubertaxi</td>\n",
       "      <td>6.55</td>\n",
       "      <td>1967.0</td>\n",
       "      <td>320.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>238256</th>\n",
       "      <td>309524</td>\n",
       "      <td>Uber</td>\n",
       "      <td>basic</td>\n",
       "      <td>uberx</td>\n",
       "      <td>6.55</td>\n",
       "      <td>1926.0</td>\n",
       "      <td>338.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>238257</th>\n",
       "      <td>309524</td>\n",
       "      <td>Uber</td>\n",
       "      <td>premium</td>\n",
       "      <td>uberselect</td>\n",
       "      <td>12.66</td>\n",
       "      <td>1926.0</td>\n",
       "      <td>338.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>238260</th>\n",
       "      <td>309524</td>\n",
       "      <td>Uber</td>\n",
       "      <td>taxi</td>\n",
       "      <td>ubertaxi</td>\n",
       "      <td>6.55</td>\n",
       "      <td>1926.0</td>\n",
       "      <td>338.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>346699</th>\n",
       "      <td>309525</td>\n",
       "      <td>Uber</td>\n",
       "      <td>basic</td>\n",
       "      <td>uberx</td>\n",
       "      <td>10.27</td>\n",
       "      <td>5554.0</td>\n",
       "      <td>749.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>346703</th>\n",
       "      <td>309525</td>\n",
       "      <td>Uber</td>\n",
       "      <td>premium</td>\n",
       "      <td>uberselect</td>\n",
       "      <td>22.28</td>\n",
       "      <td>5554.0</td>\n",
       "      <td>749.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>283713</th>\n",
       "      <td>445780</td>\n",
       "      <td>Uber</td>\n",
       "      <td>pool</td>\n",
       "      <td>uberpool</td>\n",
       "      <td>7.94</td>\n",
       "      <td>2787.0</td>\n",
       "      <td>653.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>283718</th>\n",
       "      <td>445780</td>\n",
       "      <td>Uber</td>\n",
       "      <td>premium</td>\n",
       "      <td>uberselect</td>\n",
       "      <td>16.75</td>\n",
       "      <td>2787.0</td>\n",
       "      <td>653.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>283720</th>\n",
       "      <td>445780</td>\n",
       "      <td>Uber</td>\n",
       "      <td>taxi</td>\n",
       "      <td>ubertaxi</td>\n",
       "      <td>7.94</td>\n",
       "      <td>2787.0</td>\n",
       "      <td>653.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>97860</th>\n",
       "      <td>445781</td>\n",
       "      <td>Uber</td>\n",
       "      <td>basic</td>\n",
       "      <td>uberx</td>\n",
       "      <td>14.39</td>\n",
       "      <td>8173.0</td>\n",
       "      <td>1364.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>97869</th>\n",
       "      <td>445781</td>\n",
       "      <td>Uber</td>\n",
       "      <td>other</td>\n",
       "      <td>assist</td>\n",
       "      <td>14.39</td>\n",
       "      <td>8173.0</td>\n",
       "      <td>1364.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>97861</th>\n",
       "      <td>445781</td>\n",
       "      <td>Uber</td>\n",
       "      <td>pool</td>\n",
       "      <td>uberpool</td>\n",
       "      <td>14.39</td>\n",
       "      <td>8173.0</td>\n",
       "      <td>1364.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>97864</th>\n",
       "      <td>445781</td>\n",
       "      <td>Uber</td>\n",
       "      <td>premium</td>\n",
       "      <td>uberselect</td>\n",
       "      <td>31.88</td>\n",
       "      <td>8173.0</td>\n",
       "      <td>1364.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>97863</th>\n",
       "      <td>445781</td>\n",
       "      <td>Uber</td>\n",
       "      <td>taxi</td>\n",
       "      <td>ubertaxi</td>\n",
       "      <td>14.39</td>\n",
       "      <td>8173.0</td>\n",
       "      <td>1364.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>196497</th>\n",
       "      <td>445782</td>\n",
       "      <td>Uber</td>\n",
       "      <td>basic</td>\n",
       "      <td>uberx</td>\n",
       "      <td>10.02</td>\n",
       "      <td>4426.0</td>\n",
       "      <td>903.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>196496</th>\n",
       "      <td>445782</td>\n",
       "      <td>Uber</td>\n",
       "      <td>other</td>\n",
       "      <td>assist</td>\n",
       "      <td>10.02</td>\n",
       "      <td>4426.0</td>\n",
       "      <td>903.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>196489</th>\n",
       "      <td>445782</td>\n",
       "      <td>Uber</td>\n",
       "      <td>pool</td>\n",
       "      <td>uberpool</td>\n",
       "      <td>10.02</td>\n",
       "      <td>4426.0</td>\n",
       "      <td>903.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>196493</th>\n",
       "      <td>445782</td>\n",
       "      <td>Uber</td>\n",
       "      <td>premium</td>\n",
       "      <td>uberselect</td>\n",
       "      <td>21.64</td>\n",
       "      <td>4426.0</td>\n",
       "      <td>903.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>196490</th>\n",
       "      <td>445782</td>\n",
       "      <td>Uber</td>\n",
       "      <td>taxi</td>\n",
       "      <td>ubertaxi</td>\n",
       "      <td>10.02</td>\n",
       "      <td>4426.0</td>\n",
       "      <td>903.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>154892</th>\n",
       "      <td>445790</td>\n",
       "      <td>Uber</td>\n",
       "      <td>basic</td>\n",
       "      <td>uberx</td>\n",
       "      <td>19.74</td>\n",
       "      <td>13352.0</td>\n",
       "      <td>1813.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>154904</th>\n",
       "      <td>445790</td>\n",
       "      <td>Uber</td>\n",
       "      <td>other</td>\n",
       "      <td>assist</td>\n",
       "      <td>19.74</td>\n",
       "      <td>13352.0</td>\n",
       "      <td>1813.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>154891</th>\n",
       "      <td>445790</td>\n",
       "      <td>Uber</td>\n",
       "      <td>pool</td>\n",
       "      <td>uberpool</td>\n",
       "      <td>19.74</td>\n",
       "      <td>13352.0</td>\n",
       "      <td>1813.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>154898</th>\n",
       "      <td>445790</td>\n",
       "      <td>Uber</td>\n",
       "      <td>premium</td>\n",
       "      <td>uberselect</td>\n",
       "      <td>44.47</td>\n",
       "      <td>13352.0</td>\n",
       "      <td>1813.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>129398</th>\n",
       "      <td>445791</td>\n",
       "      <td>Uber</td>\n",
       "      <td>basic</td>\n",
       "      <td>uberx</td>\n",
       "      <td>10.56</td>\n",
       "      <td>3931.0</td>\n",
       "      <td>1147.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>129410</th>\n",
       "      <td>445791</td>\n",
       "      <td>Uber</td>\n",
       "      <td>other</td>\n",
       "      <td>assist</td>\n",
       "      <td>10.56</td>\n",
       "      <td>3931.0</td>\n",
       "      <td>1147.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>129400</th>\n",
       "      <td>445791</td>\n",
       "      <td>Uber</td>\n",
       "      <td>pool</td>\n",
       "      <td>uberpool</td>\n",
       "      <td>10.56</td>\n",
       "      <td>3931.0</td>\n",
       "      <td>1147.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>129405</th>\n",
       "      <td>445791</td>\n",
       "      <td>Uber</td>\n",
       "      <td>premium</td>\n",
       "      <td>uberselect</td>\n",
       "      <td>22.83</td>\n",
       "      <td>3931.0</td>\n",
       "      <td>1147.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>129411</th>\n",
       "      <td>445791</td>\n",
       "      <td>Uber</td>\n",
       "      <td>taxi</td>\n",
       "      <td>ubertaxi</td>\n",
       "      <td>10.56</td>\n",
       "      <td>3931.0</td>\n",
       "      <td>1147.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>143955</th>\n",
       "      <td>445799</td>\n",
       "      <td>Uber</td>\n",
       "      <td>basic</td>\n",
       "      <td>uberx</td>\n",
       "      <td>8.24</td>\n",
       "      <td>2627.0</td>\n",
       "      <td>767.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>143958</th>\n",
       "      <td>445799</td>\n",
       "      <td>Uber</td>\n",
       "      <td>other</td>\n",
       "      <td>assist</td>\n",
       "      <td>8.24</td>\n",
       "      <td>2627.0</td>\n",
       "      <td>767.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>143952</th>\n",
       "      <td>445799</td>\n",
       "      <td>Uber</td>\n",
       "      <td>pool</td>\n",
       "      <td>uberpool</td>\n",
       "      <td>8.24</td>\n",
       "      <td>2627.0</td>\n",
       "      <td>767.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>143946</th>\n",
       "      <td>445799</td>\n",
       "      <td>Uber</td>\n",
       "      <td>premium</td>\n",
       "      <td>uberselect</td>\n",
       "      <td>17.43</td>\n",
       "      <td>2627.0</td>\n",
       "      <td>767.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>143954</th>\n",
       "      <td>445799</td>\n",
       "      <td>Uber</td>\n",
       "      <td>taxi</td>\n",
       "      <td>ubertaxi</td>\n",
       "      <td>8.24</td>\n",
       "      <td>2627.0</td>\n",
       "      <td>767.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>245860</th>\n",
       "      <td>445816</td>\n",
       "      <td>Uber</td>\n",
       "      <td>basic</td>\n",
       "      <td>uberx</td>\n",
       "      <td>24.91</td>\n",
       "      <td>19274.0</td>\n",
       "      <td>2070.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>245854</th>\n",
       "      <td>445816</td>\n",
       "      <td>Uber</td>\n",
       "      <td>pool</td>\n",
       "      <td>uberpool</td>\n",
       "      <td>24.91</td>\n",
       "      <td>19274.0</td>\n",
       "      <td>2070.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>245863</th>\n",
       "      <td>445816</td>\n",
       "      <td>Uber</td>\n",
       "      <td>premium</td>\n",
       "      <td>uberselect</td>\n",
       "      <td>56.73</td>\n",
       "      <td>19274.0</td>\n",
       "      <td>2070.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>123449 rows × 7 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "        search_id travel_provider uber_mode_class travel_product  \\\n",
       "224920     309474            Uber           basic          uberx   \n",
       "224921     309474            Uber         premium     uberselect   \n",
       "224925     309474            Uber            taxi       ubertaxi   \n",
       "100855     309475            Uber           basic          uberx   \n",
       "100856     309475            Uber         premium     uberselect   \n",
       "78546      309477            Uber           basic          uberx   \n",
       "78547      309477            Uber         premium     uberselect   \n",
       "59904      309481            Uber           basic          uberx   \n",
       "59907      309481            Uber         premium     uberselect   \n",
       "310828     309487            Uber           basic          uberx   \n",
       "310830     309487            Uber         premium     uberselect   \n",
       "310834     309487            Uber            taxi       ubertaxi   \n",
       "182706     309488            Uber           basic          uberx   \n",
       "182708     309488            Uber         premium     uberselect   \n",
       "182704     309488            Uber            taxi       ubertaxi   \n",
       "310837     309490            Uber           basic          uberx   \n",
       "310840     309490            Uber         premium     uberselect   \n",
       "164320     309505            Uber           basic          uberx   \n",
       "164323     309505            Uber         premium     uberselect   \n",
       "264696     309512            Uber           basic          uberx   \n",
       "264699     309512            Uber         premium     uberselect   \n",
       "264692     309512            Uber            taxi       ubertaxi   \n",
       "273201     309516            Uber           basic          uberx   \n",
       "273197     309516            Uber         premium     uberselect   \n",
       "273202     309516            Uber            taxi       ubertaxi   \n",
       "238256     309524            Uber           basic          uberx   \n",
       "238257     309524            Uber         premium     uberselect   \n",
       "238260     309524            Uber            taxi       ubertaxi   \n",
       "346699     309525            Uber           basic          uberx   \n",
       "346703     309525            Uber         premium     uberselect   \n",
       "...           ...             ...             ...            ...   \n",
       "283713     445780            Uber            pool       uberpool   \n",
       "283718     445780            Uber         premium     uberselect   \n",
       "283720     445780            Uber            taxi       ubertaxi   \n",
       "97860      445781            Uber           basic          uberx   \n",
       "97869      445781            Uber           other         assist   \n",
       "97861      445781            Uber            pool       uberpool   \n",
       "97864      445781            Uber         premium     uberselect   \n",
       "97863      445781            Uber            taxi       ubertaxi   \n",
       "196497     445782            Uber           basic          uberx   \n",
       "196496     445782            Uber           other         assist   \n",
       "196489     445782            Uber            pool       uberpool   \n",
       "196493     445782            Uber         premium     uberselect   \n",
       "196490     445782            Uber            taxi       ubertaxi   \n",
       "154892     445790            Uber           basic          uberx   \n",
       "154904     445790            Uber           other         assist   \n",
       "154891     445790            Uber            pool       uberpool   \n",
       "154898     445790            Uber         premium     uberselect   \n",
       "129398     445791            Uber           basic          uberx   \n",
       "129410     445791            Uber           other         assist   \n",
       "129400     445791            Uber            pool       uberpool   \n",
       "129405     445791            Uber         premium     uberselect   \n",
       "129411     445791            Uber            taxi       ubertaxi   \n",
       "143955     445799            Uber           basic          uberx   \n",
       "143958     445799            Uber           other         assist   \n",
       "143952     445799            Uber            pool       uberpool   \n",
       "143946     445799            Uber         premium     uberselect   \n",
       "143954     445799            Uber            taxi       ubertaxi   \n",
       "245860     445816            Uber           basic          uberx   \n",
       "245854     445816            Uber            pool       uberpool   \n",
       "245863     445816            Uber         premium     uberselect   \n",
       "\n",
       "        travel_price  travel_distance  travel_time  \n",
       "224920          6.55           2144.0        393.0  \n",
       "224921         13.49           2144.0        393.0  \n",
       "224925          6.55           2144.0        393.0  \n",
       "100855          7.48           3335.0        422.0  \n",
       "100856         15.77           3335.0        422.0  \n",
       "78546          38.95          39564.0       1945.0  \n",
       "78547          90.36          39564.0       1945.0  \n",
       "59904          22.74          20872.0       1166.0  \n",
       "59907          51.93          20872.0       1166.0  \n",
       "310828         15.29          10647.0       1128.0  \n",
       "310830         34.14          10647.0       1128.0  \n",
       "310834         15.29          10647.0       1128.0  \n",
       "182706          8.63           3521.0        699.0  \n",
       "182708         18.39           3521.0        699.0  \n",
       "182704          8.63           3521.0        699.0  \n",
       "310837         15.30          10636.0       1132.0  \n",
       "310840         34.16          10636.0       1132.0  \n",
       "164320          8.07           3859.0        481.0  \n",
       "164323         17.15           3859.0        481.0  \n",
       "264696         12.17           7008.0        984.0  \n",
       "264699         26.73           7008.0        984.0  \n",
       "264692         12.17           7008.0        984.0  \n",
       "273201          6.55           1967.0        320.0  \n",
       "273197         12.58           1967.0        320.0  \n",
       "273202          6.55           1967.0        320.0  \n",
       "238256          6.55           1926.0        338.0  \n",
       "238257         12.66           1926.0        338.0  \n",
       "238260          6.55           1926.0        338.0  \n",
       "346699         10.27           5554.0        749.0  \n",
       "346703         22.28           5554.0        749.0  \n",
       "...              ...              ...          ...  \n",
       "283713          7.94           2787.0        653.0  \n",
       "283718         16.75           2787.0        653.0  \n",
       "283720          7.94           2787.0        653.0  \n",
       "97860          14.39           8173.0       1364.0  \n",
       "97869          14.39           8173.0       1364.0  \n",
       "97861          14.39           8173.0       1364.0  \n",
       "97864          31.88           8173.0       1364.0  \n",
       "97863          14.39           8173.0       1364.0  \n",
       "196497         10.02           4426.0        903.0  \n",
       "196496         10.02           4426.0        903.0  \n",
       "196489         10.02           4426.0        903.0  \n",
       "196493         21.64           4426.0        903.0  \n",
       "196490         10.02           4426.0        903.0  \n",
       "154892         19.74          13352.0       1813.0  \n",
       "154904         19.74          13352.0       1813.0  \n",
       "154891         19.74          13352.0       1813.0  \n",
       "154898         44.47          13352.0       1813.0  \n",
       "129398         10.56           3931.0       1147.0  \n",
       "129410         10.56           3931.0       1147.0  \n",
       "129400         10.56           3931.0       1147.0  \n",
       "129405         22.83           3931.0       1147.0  \n",
       "129411         10.56           3931.0       1147.0  \n",
       "143955          8.24           2627.0        767.0  \n",
       "143958          8.24           2627.0        767.0  \n",
       "143952          8.24           2627.0        767.0  \n",
       "143946         17.43           2627.0        767.0  \n",
       "143954          8.24           2627.0        767.0  \n",
       "245860         24.91          19274.0       2070.0  \n",
       "245854         24.91          19274.0       2070.0  \n",
       "245863         56.73          19274.0       2070.0  \n",
       "\n",
       "[123449 rows x 7 columns]"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "best_uber"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "estimation_set['uber_basic_cost'] = price['basic']\n",
    "estimation_set['uber_basic_dist'] = dist['basic'] / 5280.0\n",
    "estimation_set['uber_basic_time'] = time['basic'] / 60.0\n",
    "estimation_set['uber_premium_cost'] = price['premium']\n",
    "estimation_set['uber_premium_dist'] = dist['premium'] / 5280.0\n",
    "estimation_set['uber_premium_time'] = time['premium'] / 60.0\n",
    "estimation_set['uber_taxi_cost'] = price['taxi']\n",
    "estimation_set['uber_taxi_dist'] = dist['taxi'] / 5280.0\n",
    "estimation_set['uber_taxi_time'] = time['taxi'] / 60.0\n",
    "estimation_set['uber_other_cost'] = price['other']\n",
    "estimation_set['uber_other_dist'] = dist['other'] / 5280.0\n",
    "estimation_set['uber_other_time'] = time['other'] / 60.0\n",
    "estimation_set['timestamp'] = choices.groupby('search_id')['timestamp'].first()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "dti = pd.DatetimeIndex(estimation_set['timestamp'])\n",
    "dti = dti.tz_localize('UTC')\n",
    "dti_uspac =dti.tz_convert('US/Pacific')\n",
    "estimation_set['dayofweek'] = dti_uspac.dayofweek\n",
    "estimation_set['time'] = dti_uspac.time\n",
    "estimation_set['date'] = dti_uspac.date\n",
    "estimation_set['hour'] = dti_uspac.hour\n",
    "estimation_set['late_night'] = 0\n",
    "estimation_set.loc[estimation_set['hour'].isin([22,23,0,1,2,3]),'late_night'] = 1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "estimation_subset = estimation_set[estimation_set['chosen_mode3'].isin(['transit','uber'])]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "chosen_mode\n",
      "basic        255\n",
      "premium      202\n",
      "transit    19189\n",
      "dtype: int64\n",
      "chosen_mode3\n",
      "transit    19189\n",
      "uber         457\n",
      "dtype: int64\n"
     ]
    }
   ],
   "source": [
    "print estimation_subset.groupby('chosen_mode').count().max(axis=1)\n",
    "print estimation_subset.groupby(['chosen_mode3']).count().max(axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {
    "collapsed": false,
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Python27_64\\lib\\site-packages\\ipykernel-4.3.1-py2.7.egg\\ipykernel\\__main__.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",
      "  from ipykernel import kernelapp as app\n",
      "C:\\Python27_64\\lib\\site-packages\\ipykernel-4.3.1-py2.7.egg\\ipykernel\\__main__.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",
      "  app.launch_new_instance()\n",
      "C:\\Python27_64\\lib\\site-packages\\ipykernel-4.3.1-py2.7.egg\\ipykernel\\__main__.py:4: 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",
      "C:\\Python27_64\\lib\\site-packages\\ipykernel-4.3.1-py2.7.egg\\ipykernel\\__main__.py:5: 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",
      "C:\\Python27_64\\lib\\site-packages\\ipykernel-4.3.1-py2.7.egg\\ipykernel\\__main__.py:6: 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",
      "C:\\Python27_64\\lib\\site-packages\\ipykernel-4.3.1-py2.7.egg\\ipykernel\\__main__.py:7: 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",
      "C:\\Python27_64\\lib\\site-packages\\ipykernel-4.3.1-py2.7.egg\\ipykernel\\__main__.py:8: 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",
      "C:\\Python27_64\\lib\\site-packages\\ipykernel-4.3.1-py2.7.egg\\ipykernel\\__main__.py:9: 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",
      "C:\\Python27_64\\lib\\site-packages\\ipykernel-4.3.1-py2.7.egg\\ipykernel\\__main__.py:10: 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",
      "C:\\Python27_64\\lib\\site-packages\\ipykernel-4.3.1-py2.7.egg\\ipykernel\\__main__.py:11: 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",
      "C:\\Python27_64\\lib\\site-packages\\ipykernel-4.3.1-py2.7.egg\\ipykernel\\__main__.py:12: 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",
      "C:\\Python27_64\\lib\\site-packages\\ipykernel-4.3.1-py2.7.egg\\ipykernel\\__main__.py:13: 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",
      "C:\\Python27_64\\lib\\site-packages\\pandas-0.18.1-py2.7-win-amd64.egg\\pandas\\core\\indexing.py:465: 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",
      "  self.obj[item] = s\n",
      "C:\\Python27_64\\lib\\site-packages\\ipykernel-4.3.1-py2.7.egg\\ipykernel\\__main__.py:16: 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"
     ]
    }
   ],
   "source": [
    "subset2 = estimation_set[estimation_set['chosen_mode3'].isin(['transit','uber'])]\n",
    "subset2['walkwait'] = subset2['walk'] + subset2['wait']\n",
    "subset2['ln_wait'] = np.log(subset2['wait']+1)\n",
    "subset2['ln_walk'] = np.log(subset2['walk']+1)\n",
    "subset2['ln_walkwait'] = np.log(subset2['walkwait']+1)\n",
    "subset2['ln_trn_cost'] = np.log(subset2['transit_cost']+1)\n",
    "subset2['ln_transit_time'] = np.log(subset2['transit_time']+1)\n",
    "subset2['ln_uber_basic_time'] = np.log(subset2['uber_basic_time']+1)\n",
    "subset2['ln_headway'] = np.log(subset2['headway']+1)\n",
    "subset2['tot_trn_time'] = subset2['walk'] + subset2['wait'] + subset2['transit_time']\n",
    "subset2['uber_time_save'] = subset2['tot_trn_time'] - subset2['uber_basic_time']\n",
    "subset2['uber_pct_time_save'] = subset2['uber_time_save'] / subset2['tot_trn_time']\n",
    "subset2['start_or_end_near_home'] = np.nan\n",
    "subset2.loc[(subset2['start_is_near_home']==0) & (subset2['end_is_near_home']==0),'start_or_end_near_home'] = 0\n",
    "subset2.loc[(subset2['start_is_near_home']==1) | (subset2['end_is_near_home']==1),'start_or_end_near_home'] = 1\n",
    "subset2['start_or_end_near_work'] = np.nan\n",
    "subset2.loc[(subset2['start_is_near_work']==0) & (subset2['end_is_near_work']==0),'start_or_end_near_work'] = 0\n",
    "subset2.loc[(subset2['start_is_near_work']==1) | (subset2['end_is_near_work']==1),'start_or_end_near_work'] = 1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Optimization terminated successfully.\n",
      "         Current function value: 0.105631\n",
      "         Iterations 8\n",
      "                          MNLogit Regression Results                          \n",
      "==============================================================================\n",
      "Dep. Variable:           chosen_mode3   No. Observations:                19641\n",
      "Model:                        MNLogit   Df Residuals:                    19631\n",
      "Method:                           MLE   Df Model:                            9\n",
      "Date:                Wed, 20 Jul 2016   Pseudo R-squ.:                 0.04404\n",
      "Time:                        17:55:32   Log-Likelihood:                -2074.7\n",
      "converged:                       True   LL-Null:                       -2170.3\n",
      "                                        LLR p-value:                 2.363e-36\n",
      "=====================================================================================\n",
      "chosen_mode3=uber       coef    std err          z      P>|z|      [95.0% Conf. Int.]\n",
      "-------------------------------------------------------------------------------------\n",
      "n_transit            -0.5788      0.054    -10.705      0.000        -0.685    -0.473\n",
      "n_arr                 0.0629      0.202      0.311      0.756        -0.333     0.459\n",
      "walk                 -0.0040      0.013     -0.316      0.752        -0.029     0.021\n",
      "wait                  0.0006      0.000      1.992      0.046      8.97e-06     0.001\n",
      "headway               0.0030      0.005      0.641      0.521        -0.006     0.012\n",
      "transit_time          0.0044      0.004      1.101      0.271        -0.003     0.012\n",
      "uber_basic_time       0.0095      0.009      1.058      0.290        -0.008     0.027\n",
      "transit_cost          0.0585      0.029      2.012      0.044         0.002     0.115\n",
      "uber_basic_cost      -0.0161      0.013     -1.274      0.203        -0.041     0.009\n",
      "const                -2.9183      0.646     -4.515      0.000        -4.185    -1.651\n",
      "=====================================================================================\n"
     ]
    }
   ],
   "source": [
    "# --------------------------\n",
    "# - SERIES 1               - \n",
    "# --------------------------\n",
    "# try everything I think might matter except late_night\n",
    "# n_transit: Number of transit options presented to the user\n",
    "# n_arr:     Number of arrivals for the best transit option\n",
    "# walk:      Walk time (in minutes) of the best transit option\n",
    "# wait:      Wait time (in minutes) of the best transit option for the first arrival that the user has enough time to walk to\n",
    "# headway:   Headway estimate based on 3 presented arrival times\n",
    "# transit_time: IVT for best transit option\n",
    "# uber_basic_time: IVT for best uberx option (same travel time for any uber service...)\n",
    "# transit_cost: Fare (USD) for entire transit trip (includes transfers)\n",
    "# uber_basic_cost: Fare (USD) for uberx (other uber service prices vary, does not include surge)\n",
    "\n",
    "exog_cols = ['n_transit','n_arr','walk','wait','headway','transit_time','uber_basic_time','transit_cost','uber_basic_cost']\n",
    "defn = subset2.dropna(subset=exog_cols)\n",
    "y = defn['chosen_mode3']\n",
    "x = defn.loc[:,exog_cols]\n",
    "x['const'] = 1\n",
    "\n",
    "s2_mdl = st.MNLogit(y, x)\n",
    "s2_fit = s2_mdl.fit(maxiter=100)\n",
    "print s2_fit.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Optimization terminated successfully.\n",
      "         Current function value: 0.105634\n",
      "         Iterations 8\n",
      "                          MNLogit Regression Results                          \n",
      "==============================================================================\n",
      "Dep. Variable:           chosen_mode3   No. Observations:                19641\n",
      "Model:                        MNLogit   Df Residuals:                    19632\n",
      "Method:                           MLE   Df Model:                            8\n",
      "Date:                Wed, 20 Jul 2016   Pseudo R-squ.:                 0.04402\n",
      "Time:                        17:55:32   Log-Likelihood:                -2074.8\n",
      "converged:                       True   LL-Null:                       -2170.3\n",
      "                                        LLR p-value:                 4.887e-37\n",
      "=====================================================================================\n",
      "chosen_mode3=uber       coef    std err          z      P>|z|      [95.0% Conf. Int.]\n",
      "-------------------------------------------------------------------------------------\n",
      "n_transit            -0.5794      0.054    -10.719      0.000        -0.685    -0.473\n",
      "walk                 -0.0038      0.013     -0.302      0.762        -0.029     0.021\n",
      "wait                  0.0005      0.000      2.653      0.008         0.000     0.001\n",
      "headway               0.0017      0.002      0.830      0.406        -0.002     0.006\n",
      "transit_time          0.0044      0.004      1.099      0.272        -0.003     0.012\n",
      "uber_basic_time       0.0096      0.009      1.072      0.284        -0.008     0.027\n",
      "transit_cost          0.0584      0.029      2.010      0.044         0.001     0.115\n",
      "uber_basic_cost      -0.0161      0.013     -1.277      0.202        -0.041     0.009\n",
      "const                -2.7244      0.171    -15.967      0.000        -3.059    -2.390\n",
      "=====================================================================================\n"
     ]
    }
   ],
   "source": [
    "# --------------------------\n",
    "# - SERIES 1               - \n",
    "# --------------------------\n",
    "# drop n_arr\n",
    "exog_cols = ['n_transit','walk','wait','headway','transit_time','uber_basic_time','transit_cost','uber_basic_cost']\n",
    "defn = subset2.dropna(subset=exog_cols)\n",
    "y = defn['chosen_mode3']\n",
    "x = defn.loc[:,exog_cols]\n",
    "x['const'] = 1\n",
    "\n",
    "s2_mdl = st.MNLogit(y, x)\n",
    "s2_fit = s2_mdl.fit(maxiter=100)\n",
    "print s2_fit.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Optimization terminated successfully.\n",
      "         Current function value: 0.105636\n",
      "         Iterations 8\n",
      "                          MNLogit Regression Results                          \n",
      "==============================================================================\n",
      "Dep. Variable:           chosen_mode3   No. Observations:                19641\n",
      "Model:                        MNLogit   Df Residuals:                    19633\n",
      "Method:                           MLE   Df Model:                            7\n",
      "Date:                Wed, 20 Jul 2016   Pseudo R-squ.:                 0.04399\n",
      "Time:                        17:55:32   Log-Likelihood:                -2074.8\n",
      "converged:                       True   LL-Null:                       -2170.3\n",
      "                                        LLR p-value:                 9.391e-38\n",
      "=====================================================================================\n",
      "chosen_mode3=uber       coef    std err          z      P>|z|      [95.0% Conf. Int.]\n",
      "-------------------------------------------------------------------------------------\n",
      "n_transit            -0.5804      0.054    -10.763      0.000        -0.686    -0.475\n",
      "wait                  0.0005      0.000      2.666      0.008         0.000     0.001\n",
      "headway               0.0017      0.002      0.816      0.415        -0.002     0.006\n",
      "transit_time          0.0041      0.004      1.054      0.292        -0.003     0.012\n",
      "uber_basic_time       0.0097      0.009      1.076      0.282        -0.008     0.027\n",
      "transit_cost          0.0589      0.029      2.030      0.042         0.002     0.116\n",
      "uber_basic_cost      -0.0160      0.013     -1.265      0.206        -0.041     0.009\n",
      "const                -2.7270      0.170    -15.996      0.000        -3.061    -2.393\n",
      "=====================================================================================\n"
     ]
    }
   ],
   "source": [
    "# --------------------------\n",
    "# - SERIES 1               - \n",
    "# --------------------------\n",
    "# drop walk\n",
    "exog_cols = ['n_transit','wait','headway','transit_time','uber_basic_time','transit_cost','uber_basic_cost']\n",
    "defn = subset2.dropna(subset=exog_cols)\n",
    "y = defn['chosen_mode3']\n",
    "x = defn.loc[:,exog_cols]\n",
    "x['const'] = 1\n",
    "\n",
    "s2_mdl = st.MNLogit(y, x)\n",
    "s2_fit = s2_mdl.fit(maxiter=100)\n",
    "print s2_fit.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Optimization terminated successfully.\n",
      "         Current function value: 0.105652\n",
      "         Iterations 8\n",
      "                          MNLogit Regression Results                          \n",
      "==============================================================================\n",
      "Dep. Variable:           chosen_mode3   No. Observations:                19641\n",
      "Model:                        MNLogit   Df Residuals:                    19634\n",
      "Method:                           MLE   Df Model:                            6\n",
      "Date:                Wed, 20 Jul 2016   Pseudo R-squ.:                 0.04385\n",
      "Time:                        17:55:32   Log-Likelihood:                -2075.1\n",
      "converged:                       True   LL-Null:                       -2170.3\n",
      "                                        LLR p-value:                 2.180e-38\n",
      "=====================================================================================\n",
      "chosen_mode3=uber       coef    std err          z      P>|z|      [95.0% Conf. Int.]\n",
      "-------------------------------------------------------------------------------------\n",
      "n_transit            -0.5801      0.054    -10.754      0.000        -0.686    -0.474\n",
      "wait                  0.0006   9.89e-05      6.241      0.000         0.000     0.001\n",
      "transit_time          0.0041      0.004      1.050      0.294        -0.004     0.012\n",
      "uber_basic_time       0.0090      0.009      1.003      0.316        -0.009     0.026\n",
      "transit_cost          0.0606      0.029      2.088      0.037         0.004     0.118\n",
      "uber_basic_cost      -0.0153      0.013     -1.217      0.224        -0.040     0.009\n",
      "const                -2.6928      0.165    -16.331      0.000        -3.016    -2.370\n",
      "=====================================================================================\n"
     ]
    }
   ],
   "source": [
    "# --------------------------\n",
    "# - SERIES 1               - \n",
    "# --------------------------\n",
    "# drop headway\n",
    "exog_cols = ['n_transit','wait','transit_time','uber_basic_time','transit_cost','uber_basic_cost']\n",
    "defn = subset2.dropna(subset=exog_cols)\n",
    "y = defn['chosen_mode3']\n",
    "x = defn.loc[:,exog_cols]\n",
    "x['const'] = 1\n",
    "\n",
    "s2_mdl = st.MNLogit(y, x)\n",
    "s2_fit = s2_mdl.fit(maxiter=100)\n",
    "print s2_fit.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Optimization terminated successfully.\n",
      "         Current function value: 0.105678\n",
      "         Iterations 8\n",
      "                          MNLogit Regression Results                          \n",
      "==============================================================================\n",
      "Dep. Variable:           chosen_mode3   No. Observations:                19641\n",
      "Model:                        MNLogit   Df Residuals:                    19635\n",
      "Method:                           MLE   Df Model:                            5\n",
      "Date:                Wed, 20 Jul 2016   Pseudo R-squ.:                 0.04362\n",
      "Time:                        17:55:33   Log-Likelihood:                -2075.6\n",
      "converged:                       True   LL-Null:                       -2170.3\n",
      "                                        LLR p-value:                 5.468e-39\n",
      "=====================================================================================\n",
      "chosen_mode3=uber       coef    std err          z      P>|z|      [95.0% Conf. Int.]\n",
      "-------------------------------------------------------------------------------------\n",
      "n_transit            -0.5738      0.053    -10.730      0.000        -0.679    -0.469\n",
      "wait                  0.0006    9.9e-05      6.221      0.000         0.000     0.001\n",
      "transit_time          0.0039      0.004      1.021      0.307        -0.004     0.011\n",
      "transit_cost          0.0556      0.029      1.933      0.053        -0.001     0.112\n",
      "uber_basic_cost      -0.0054      0.008     -0.709      0.478        -0.020     0.010\n",
      "const                -2.6671      0.162    -16.467      0.000        -2.985    -2.350\n",
      "=====================================================================================\n"
     ]
    }
   ],
   "source": [
    "# --------------------------\n",
    "# - SERIES 1               - \n",
    "# --------------------------\n",
    "# drop uber time\n",
    "exog_cols = ['n_transit','wait','transit_time','transit_cost','uber_basic_cost']\n",
    "defn = subset2.dropna(subset=exog_cols)\n",
    "y = defn['chosen_mode3']\n",
    "x = defn.loc[:,exog_cols]\n",
    "x['const'] = 1\n",
    "\n",
    "s2_mdl = st.MNLogit(y, x)\n",
    "s2_fit = s2_mdl.fit(maxiter=100)\n",
    "print s2_fit.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Optimization terminated successfully.\n",
      "         Current function value: 0.105683\n",
      "         Iterations 8\n",
      "                          MNLogit Regression Results                          \n",
      "==============================================================================\n",
      "Dep. Variable:           chosen_mode3   No. Observations:                19646\n",
      "Model:                        MNLogit   Df Residuals:                    19642\n",
      "Method:                           MLE   Df Model:                            3\n",
      "Date:                Wed, 20 Jul 2016   Pseudo R-squ.:                 0.04338\n",
      "Time:                        17:55:33   Log-Likelihood:                -2076.2\n",
      "converged:                       True   LL-Null:                       -2170.4\n",
      "                                        LLR p-value:                 1.420e-40\n",
      "=====================================================================================\n",
      "chosen_mode3=uber       coef    std err          z      P>|z|      [95.0% Conf. Int.]\n",
      "-------------------------------------------------------------------------------------\n",
      "n_transit            -0.5690      0.053    -10.787      0.000        -0.672    -0.466\n",
      "wait                  0.0006   9.89e-05      6.234      0.000         0.000     0.001\n",
      "transit_cost          0.0610      0.020      3.053      0.002         0.022     0.100\n",
      "const                -2.6575      0.158    -16.807      0.000        -2.967    -2.348\n",
      "=====================================================================================\n"
     ]
    }
   ],
   "source": [
    "# --------------------------\n",
    "# - SERIES 1               - \n",
    "# --------------------------\n",
    "# drop transit_time and uber_cost\n",
    "exog_cols = ['n_transit','wait','transit_cost']\n",
    "defn = subset2.dropna(subset=exog_cols)\n",
    "y = defn['chosen_mode3']\n",
    "x = defn.loc[:,exog_cols]\n",
    "x['const'] = 1\n",
    "\n",
    "s2_mdl = st.MNLogit(y, x)\n",
    "s2_fit = s2_mdl.fit(maxiter=100)\n",
    "print s2_fit.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Optimization terminated successfully.\n",
      "         Current function value: 0.106525\n",
      "         Iterations 8\n",
      "                          MNLogit Regression Results                          \n",
      "==============================================================================\n",
      "Dep. Variable:           chosen_mode3   No. Observations:                16026\n",
      "Model:                        MNLogit   Df Residuals:                    16012\n",
      "Method:                           MLE   Df Model:                           13\n",
      "Date:                Wed, 20 Jul 2016   Pseudo R-squ.:                 0.05213\n",
      "Time:                        17:55:33   Log-Likelihood:                -1707.2\n",
      "converged:                       True   LL-Null:                       -1801.1\n",
      "                                        LLR p-value:                 4.387e-33\n",
      "======================================================================================\n",
      " chosen_mode3=uber       coef    std err          z      P>|z|      [95.0% Conf. Int.]\n",
      "--------------------------------------------------------------------------------------\n",
      "n_transit             -0.5427      0.060     -9.098      0.000        -0.660    -0.426\n",
      "n_arr                 -0.0369      0.221     -0.167      0.868        -0.471     0.397\n",
      "walk               -7.472e-05      0.015     -0.005      0.996        -0.029     0.029\n",
      "wait                   0.0006      0.000      1.873      0.061     -2.68e-05     0.001\n",
      "headway                0.0003      0.005      0.061      0.952        -0.010     0.010\n",
      "transit_time          -0.0031      0.005     -0.633      0.527        -0.013     0.007\n",
      "uber_basic_time       -0.0113      0.011     -1.070      0.284        -0.032     0.009\n",
      "transit_cost           0.0178      0.044      0.403      0.687        -0.069     0.104\n",
      "uber_basic_cost        0.0195      0.015      1.265      0.206        -0.011     0.050\n",
      "start_is_near_home     0.2301      0.136      1.694      0.090        -0.036     0.496\n",
      "end_is_near_home       0.2964      0.136      2.183      0.029         0.030     0.563\n",
      "start_is_near_work     0.3766      0.156      2.418      0.016         0.071     0.682\n",
      "end_is_near_work       0.7548      0.134      5.643      0.000         0.493     1.017\n",
      "const                 -2.8961      0.714     -4.056      0.000        -4.296    -1.497\n",
      "======================================================================================\n"
     ]
    }
   ],
   "source": [
    "# --------------------------\n",
    "# - SERIES 2               - \n",
    "# --------------------------\n",
    "# start again with everything; add is_near_home and work\n",
    "exog_cols = ['n_transit','n_arr','walk','wait','headway','transit_time',\n",
    "             'uber_basic_time','transit_cost','uber_basic_cost',\n",
    "             'start_is_near_home','end_is_near_home','start_is_near_work','end_is_near_work']\n",
    "defn = subset2.dropna(subset=exog_cols)\n",
    "y = defn['chosen_mode3']\n",
    "x = defn.loc[:,exog_cols]\n",
    "x['const'] = 1\n",
    "\n",
    "s2_mdl = st.MNLogit(y, x)\n",
    "s2_fit = s2_mdl.fit(maxiter=100)\n",
    "print s2_fit.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Optimization terminated successfully.\n",
      "         Current function value: 0.106526\n",
      "         Iterations 8\n",
      "                          MNLogit Regression Results                          \n",
      "==============================================================================\n",
      "Dep. Variable:           chosen_mode3   No. Observations:                16026\n",
      "Model:                        MNLogit   Df Residuals:                    16014\n",
      "Method:                           MLE   Df Model:                           11\n",
      "Date:                Wed, 20 Jul 2016   Pseudo R-squ.:                 0.05212\n",
      "Time:                        17:55:33   Log-Likelihood:                -1707.2\n",
      "converged:                       True   LL-Null:                       -1801.1\n",
      "                                        LLR p-value:                 2.576e-34\n",
      "======================================================================================\n",
      " chosen_mode3=uber       coef    std err          z      P>|z|      [95.0% Conf. Int.]\n",
      "--------------------------------------------------------------------------------------\n",
      "n_transit             -0.5424      0.059     -9.121      0.000        -0.659    -0.426\n",
      "wait                   0.0006      0.000      2.950      0.003         0.000     0.001\n",
      "headway                0.0011      0.002      0.464      0.642        -0.003     0.006\n",
      "transit_time          -0.0031      0.005     -0.664      0.507        -0.012     0.006\n",
      "uber_basic_time       -0.0113      0.011     -1.077      0.282        -0.032     0.009\n",
      "transit_cost           0.0179      0.044      0.405      0.685        -0.069     0.104\n",
      "uber_basic_cost        0.0195      0.015      1.267      0.205        -0.011     0.050\n",
      "start_is_near_home     0.2304      0.135      1.702      0.089        -0.035     0.496\n",
      "end_is_near_home       0.2972      0.136      2.190      0.028         0.031     0.563\n",
      "start_is_near_work     0.3760      0.156      2.416      0.016         0.071     0.681\n",
      "end_is_near_work       0.7547      0.134      5.643      0.000         0.493     1.017\n",
      "const                 -3.0100      0.211    -14.284      0.000        -3.423    -2.597\n",
      "======================================================================================\n"
     ]
    }
   ],
   "source": [
    "# --------------------------\n",
    "# - SERIES 2               - \n",
    "# --------------------------\n",
    "# drop walk and n_arr\n",
    "exog_cols = ['n_transit','wait','headway','transit_time',\n",
    "             'uber_basic_time','transit_cost','uber_basic_cost',\n",
    "             'start_is_near_home','end_is_near_home','start_is_near_work','end_is_near_work']\n",
    "defn = subset2.dropna(subset=exog_cols)\n",
    "y = defn['chosen_mode3']\n",
    "x = defn.loc[:,exog_cols]\n",
    "x['const'] = 1\n",
    "\n",
    "s2_mdl = st.MNLogit(y, x)\n",
    "s2_fit = s2_mdl.fit(maxiter=100)\n",
    "print s2_fit.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Optimization terminated successfully.\n",
      "         Current function value: 0.106550\n",
      "         Iterations 8\n",
      "                          MNLogit Regression Results                          \n",
      "==============================================================================\n",
      "Dep. Variable:           chosen_mode3   No. Observations:                16026\n",
      "Model:                        MNLogit   Df Residuals:                    16017\n",
      "Method:                           MLE   Df Model:                            8\n",
      "Date:                Wed, 20 Jul 2016   Pseudo R-squ.:                 0.05191\n",
      "Time:                        17:55:33   Log-Likelihood:                -1707.6\n",
      "converged:                       True   LL-Null:                       -1801.1\n",
      "                                        LLR p-value:                 3.524e-36\n",
      "======================================================================================\n",
      " chosen_mode3=uber       coef    std err          z      P>|z|      [95.0% Conf. Int.]\n",
      "--------------------------------------------------------------------------------------\n",
      "n_transit             -0.5437      0.059     -9.182      0.000        -0.660    -0.428\n",
      "wait                   0.0007      0.000      6.514      0.000         0.000     0.001\n",
      "uber_basic_time       -0.0125      0.010     -1.223      0.221        -0.032     0.008\n",
      "uber_basic_cost        0.0185      0.011      1.615      0.106        -0.004     0.041\n",
      "start_is_near_home     0.2131      0.133      1.602      0.109        -0.048     0.474\n",
      "end_is_near_home       0.2889      0.135      2.144      0.032         0.025     0.553\n",
      "start_is_near_work     0.3700      0.155      2.381      0.017         0.065     0.675\n",
      "end_is_near_work       0.7484      0.133      5.613      0.000         0.487     1.010\n",
      "const                 -2.9771      0.199    -14.928      0.000        -3.368    -2.586\n",
      "======================================================================================\n"
     ]
    }
   ],
   "source": [
    "# --------------------------\n",
    "# - SERIES 2               - \n",
    "# --------------------------\n",
    "# drop headway, transit_cost, transit_time\n",
    "exog_cols = ['n_transit','wait',\n",
    "             'uber_basic_time','uber_basic_cost',\n",
    "             'start_is_near_home','end_is_near_home','start_is_near_work','end_is_near_work']\n",
    "defn = subset2.dropna(subset=exog_cols)\n",
    "y = defn['chosen_mode3']\n",
    "x = defn.loc[:,exog_cols]\n",
    "x['const'] = 1\n",
    "\n",
    "s2_mdl = st.MNLogit(y, x)\n",
    "s2_fit = s2_mdl.fit(maxiter=100)\n",
    "print s2_fit.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Optimization terminated successfully.\n",
      "         Current function value: 0.106597\n",
      "         Iterations 8\n",
      "                          MNLogit Regression Results                          \n",
      "==============================================================================\n",
      "Dep. Variable:           chosen_mode3   No. Observations:                16026\n",
      "Model:                        MNLogit   Df Residuals:                    16018\n",
      "Method:                           MLE   Df Model:                            7\n",
      "Date:                Wed, 20 Jul 2016   Pseudo R-squ.:                 0.05149\n",
      "Time:                        17:55:33   Log-Likelihood:                -1708.3\n",
      "converged:                       True   LL-Null:                       -1801.1\n",
      "                                        LLR p-value:                 1.363e-36\n",
      "======================================================================================\n",
      " chosen_mode3=uber       coef    std err          z      P>|z|      [95.0% Conf. Int.]\n",
      "--------------------------------------------------------------------------------------\n",
      "n_transit             -0.5512      0.059     -9.334      0.000        -0.667    -0.435\n",
      "wait                   0.0007      0.000      6.585      0.000         0.000     0.001\n",
      "uber_basic_cost        0.0056      0.005      1.127      0.260        -0.004     0.015\n",
      "start_is_near_home     0.2033      0.133      1.531      0.126        -0.057     0.464\n",
      "end_is_near_home       0.2842      0.135      2.110      0.035         0.020     0.548\n",
      "start_is_near_work     0.3376      0.153      2.205      0.027         0.038     0.638\n",
      "end_is_near_work       0.7264      0.132      5.494      0.000         0.467     0.985\n",
      "const                 -2.9880      0.201    -14.899      0.000        -3.381    -2.595\n",
      "======================================================================================\n"
     ]
    }
   ],
   "source": [
    "# --------------------------\n",
    "# - SERIES 2               - \n",
    "# --------------------------\n",
    "# drop uber_basic_time\n",
    "exog_cols = ['n_transit','wait','uber_basic_cost',\n",
    "             'start_is_near_home','end_is_near_home','start_is_near_work','end_is_near_work']\n",
    "defn = subset2.dropna(subset=exog_cols)\n",
    "y = defn['chosen_mode3']\n",
    "x = defn.loc[:,exog_cols]\n",
    "x['const'] = 1\n",
    "\n",
    "s2_mdl = st.MNLogit(y, x)\n",
    "s2_fit = s2_mdl.fit(maxiter=100)\n",
    "print s2_fit.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Optimization terminated successfully.\n",
      "         Current function value: 0.106609\n",
      "         Iterations 8\n",
      "                          MNLogit Regression Results                          \n",
      "==============================================================================\n",
      "Dep. Variable:           chosen_mode3   No. Observations:                16031\n",
      "Model:                        MNLogit   Df Residuals:                    16024\n",
      "Method:                           MLE   Df Model:                            6\n",
      "Date:                Wed, 20 Jul 2016   Pseudo R-squ.:                 0.05115\n",
      "Time:                        17:55:34   Log-Likelihood:                -1709.0\n",
      "converged:                       True   LL-Null:                       -1801.2\n",
      "                                        LLR p-value:                 4.217e-37\n",
      "======================================================================================\n",
      " chosen_mode3=uber       coef    std err          z      P>|z|      [95.0% Conf. Int.]\n",
      "--------------------------------------------------------------------------------------\n",
      "n_transit             -0.5629      0.058     -9.684      0.000        -0.677    -0.449\n",
      "wait                   0.0007      0.000      6.725      0.000         0.001     0.001\n",
      "start_is_near_home     0.2035      0.133      1.532      0.126        -0.057     0.464\n",
      "end_is_near_home       0.2781      0.135      2.067      0.039         0.014     0.542\n",
      "start_is_near_work     0.3446      0.153      2.253      0.024         0.045     0.644\n",
      "end_is_near_work       0.7322      0.132      5.540      0.000         0.473     0.991\n",
      "const                 -2.8880      0.179    -16.164      0.000        -3.238    -2.538\n",
      "======================================================================================\n"
     ]
    }
   ],
   "source": [
    "# --------------------------\n",
    "# - SERIES 2               - \n",
    "# --------------------------\n",
    "# drop uber_basic_cost\n",
    "exog_cols = ['n_transit','wait',\n",
    "             'start_is_near_home','end_is_near_home','start_is_near_work','end_is_near_work']\n",
    "defn = subset2.dropna(subset=exog_cols)\n",
    "y = defn['chosen_mode3']\n",
    "x = defn.loc[:,exog_cols]\n",
    "x['const'] = 1\n",
    "\n",
    "s2_mdl = st.MNLogit(y, x)\n",
    "s2_fit = s2_mdl.fit(maxiter=100)\n",
    "print s2_fit.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Optimization terminated successfully.\n",
      "         Current function value: 0.106682\n",
      "         Iterations 8\n",
      "                          MNLogit Regression Results                          \n",
      "==============================================================================\n",
      "Dep. Variable:           chosen_mode3   No. Observations:                16031\n",
      "Model:                        MNLogit   Df Residuals:                    16025\n",
      "Method:                           MLE   Df Model:                            5\n",
      "Date:                Wed, 20 Jul 2016   Pseudo R-squ.:                 0.05050\n",
      "Time:                        17:55:34   Log-Likelihood:                -1710.2\n",
      "converged:                       True   LL-Null:                       -1801.2\n",
      "                                        LLR p-value:                 2.077e-37\n",
      "======================================================================================\n",
      " chosen_mode3=uber       coef    std err          z      P>|z|      [95.0% Conf. Int.]\n",
      "--------------------------------------------------------------------------------------\n",
      "n_transit             -0.5587      0.058     -9.630      0.000        -0.672    -0.445\n",
      "wait                   0.0007      0.000      6.735      0.000         0.001     0.001\n",
      "end_is_near_home       0.2132      0.127      1.685      0.092        -0.035     0.461\n",
      "start_is_near_work     0.3159      0.152      2.085      0.037         0.019     0.613\n",
      "end_is_near_work       0.7784      0.129      6.048      0.000         0.526     1.031\n",
      "const                 -2.8260      0.173    -16.309      0.000        -3.166    -2.486\n",
      "======================================================================================\n"
     ]
    }
   ],
   "source": [
    "# --------------------------\n",
    "# - SERIES 2               - \n",
    "# --------------------------\n",
    "# drop start_is_near_home\n",
    "exog_cols = ['n_transit','wait',\n",
    "             'end_is_near_home','start_is_near_work','end_is_near_work']\n",
    "defn = subset2.dropna(subset=exog_cols)\n",
    "y = defn['chosen_mode3']\n",
    "x = defn.loc[:,exog_cols]\n",
    "x['const'] = 1\n",
    "\n",
    "s2_mdl = st.MNLogit(y, x)\n",
    "s2_fit = s2_mdl.fit(maxiter=100)\n",
    "print s2_fit.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Optimization terminated successfully.\n",
      "         Current function value: 0.106315\n",
      "         Iterations 8\n",
      "                          MNLogit Regression Results                          \n",
      "==============================================================================\n",
      "Dep. Variable:           chosen_mode3   No. Observations:                16259\n",
      "Model:                        MNLogit   Df Residuals:                    16254\n",
      "Method:                           MLE   Df Model:                            4\n",
      "Date:                Wed, 20 Jul 2016   Pseudo R-squ.:                 0.04909\n",
      "Time:                        17:55:34   Log-Likelihood:                -1728.6\n",
      "converged:                       True   LL-Null:                       -1817.8\n",
      "                                        LLR p-value:                 1.592e-37\n",
      "======================================================================================\n",
      " chosen_mode3=uber       coef    std err          z      P>|z|      [95.0% Conf. Int.]\n",
      "--------------------------------------------------------------------------------------\n",
      "n_transit             -0.5597      0.058     -9.692      0.000        -0.673    -0.446\n",
      "wait                   0.0007      0.000      6.757      0.000         0.001     0.001\n",
      "start_is_near_work     0.3469      0.150      2.319      0.020         0.054     0.640\n",
      "end_is_near_work       0.6833      0.116      5.875      0.000         0.455     0.911\n",
      "const                 -2.7394      0.162    -16.896      0.000        -3.057    -2.422\n",
      "======================================================================================\n"
     ]
    }
   ],
   "source": [
    "# --------------------------\n",
    "# - SERIES 2               - \n",
    "# --------------------------\n",
    "# drop end_is_near_home\n",
    "exog_cols = ['n_transit','wait',\n",
    "             'start_is_near_work','end_is_near_work']\n",
    "defn = subset2.dropna(subset=exog_cols)\n",
    "y = defn['chosen_mode3']\n",
    "x = defn.loc[:,exog_cols]\n",
    "x['const'] = 1\n",
    "\n",
    "s2_mdl = st.MNLogit(y, x)\n",
    "s2_fit = s2_mdl.fit(maxiter=100)\n",
    "print s2_fit.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Optimization terminated successfully.\n",
      "         Current function value: 0.106676\n",
      "         Iterations 8\n",
      "                          MNLogit Regression Results                          \n",
      "==============================================================================\n",
      "Dep. Variable:           chosen_mode3   No. Observations:                16026\n",
      "Model:                        MNLogit   Df Residuals:                    16015\n",
      "Method:                           MLE   Df Model:                           10\n",
      "Date:                Wed, 20 Jul 2016   Pseudo R-squ.:                 0.05079\n",
      "Time:                        17:55:35   Log-Likelihood:                -1709.6\n",
      "converged:                       True   LL-Null:                       -1801.1\n",
      "                                        LLR p-value:                 5.710e-34\n",
      "==========================================================================================\n",
      "     chosen_mode3=uber       coef    std err          z      P>|z|      [95.0% Conf. Int.]\n",
      "------------------------------------------------------------------------------------------\n",
      "n_transit                 -0.5447      0.059     -9.174      0.000        -0.661    -0.428\n",
      "n_arr                     -0.0425      0.221     -0.193      0.847        -0.475     0.390\n",
      "walkwait                   0.0006      0.000      1.883      0.060     -2.37e-05     0.001\n",
      "headway                    0.0001      0.005      0.028      0.978        -0.010     0.010\n",
      "transit_time              -0.0030      0.005     -0.649      0.516        -0.012     0.006\n",
      "uber_basic_time           -0.0124      0.011     -1.180      0.238        -0.033     0.008\n",
      "transit_cost               0.0153      0.044      0.348      0.728        -0.071     0.102\n",
      "uber_basic_cost            0.0209      0.015      1.360      0.174        -0.009     0.051\n",
      "start_or_end_near_home     0.2656      0.111      2.386      0.017         0.047     0.484\n",
      "start_or_end_near_work     0.6084      0.108      5.638      0.000         0.397     0.820\n",
      "const                     -2.8655      0.712     -4.026      0.000        -4.261    -1.470\n",
      "==========================================================================================\n"
     ]
    }
   ],
   "source": [
    "# --------------------------\n",
    "# - SERIES 3               - \n",
    "# --------------------------\n",
    "# start again with everything including is_near_home and work.  \n",
    "# combine walk and wait to walkwait (access time to transit).\n",
    "# combine start_is_near_home and end_is_near_home into start_or_end_near_home.\n",
    "# combine start_is_near_work and end_is_near_work into start_or_end_near_work\n",
    "exog_cols = ['n_transit','n_arr','walkwait','headway','transit_time',\n",
    "             'uber_basic_time','transit_cost','uber_basic_cost',\n",
    "             'start_or_end_near_home','start_or_end_near_work']\n",
    "defn = subset2.dropna(subset=exog_cols)\n",
    "y = defn['chosen_mode3']\n",
    "x = defn.loc[:,exog_cols]\n",
    "x['const'] = 1\n",
    "\n",
    "s2_mdl = st.MNLogit(y, x)\n",
    "s2_fit = s2_mdl.fit(maxiter=100)\n",
    "print s2_fit.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Optimization terminated successfully.\n",
      "         Current function value: 0.106683\n",
      "         Iterations 8\n",
      "                          MNLogit Regression Results                          \n",
      "==============================================================================\n",
      "Dep. Variable:           chosen_mode3   No. Observations:                16026\n",
      "Model:                        MNLogit   Df Residuals:                    16017\n",
      "Method:                           MLE   Df Model:                            8\n",
      "Date:                Wed, 20 Jul 2016   Pseudo R-squ.:                 0.05073\n",
      "Time:                        17:55:35   Log-Likelihood:                -1709.7\n",
      "converged:                       True   LL-Null:                       -1801.1\n",
      "                                        LLR p-value:                 2.758e-35\n",
      "==========================================================================================\n",
      "     chosen_mode3=uber       coef    std err          z      P>|z|      [95.0% Conf. Int.]\n",
      "------------------------------------------------------------------------------------------\n",
      "n_transit                 -0.5442      0.059     -9.167      0.000        -0.661    -0.428\n",
      "walkwait                   0.0007      0.000      6.525      0.000         0.000     0.001\n",
      "transit_time              -0.0031      0.005     -0.658      0.511        -0.012     0.006\n",
      "uber_basic_time           -0.0129      0.010     -1.232      0.218        -0.033     0.008\n",
      "transit_cost               0.0166      0.044      0.375      0.707        -0.070     0.103\n",
      "uber_basic_cost            0.0213      0.015      1.392      0.164        -0.009     0.051\n",
      "start_or_end_near_home     0.2675      0.111      2.404      0.016         0.049     0.485\n",
      "start_or_end_near_work     0.6073      0.108      5.630      0.000         0.396     0.819\n",
      "const                     -2.9772      0.204    -14.601      0.000        -3.377    -2.578\n",
      "==========================================================================================\n"
     ]
    }
   ],
   "source": [
    "# --------------------------\n",
    "# - SERIES 3               - \n",
    "# --------------------------\n",
    "# drop headway, n_arr\n",
    "exog_cols = ['n_transit','walkwait','transit_time',\n",
    "             'uber_basic_time','transit_cost','uber_basic_cost',\n",
    "             'start_or_end_near_home','start_or_end_near_work']\n",
    "defn = subset2.dropna(subset=exog_cols)\n",
    "y = defn['chosen_mode3']\n",
    "x = defn.loc[:,exog_cols]\n",
    "x['const'] = 1\n",
    "\n",
    "s2_mdl = st.MNLogit(y, x)\n",
    "s2_fit = s2_mdl.fit(maxiter=100)\n",
    "print s2_fit.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Optimization terminated successfully.\n",
      "         Current function value: 0.106699\n",
      "         Iterations 8\n",
      "                          MNLogit Regression Results                          \n",
      "==============================================================================\n",
      "Dep. Variable:           chosen_mode3   No. Observations:                16026\n",
      "Model:                        MNLogit   Df Residuals:                    16019\n",
      "Method:                           MLE   Df Model:                            6\n",
      "Date:                Wed, 20 Jul 2016   Pseudo R-squ.:                 0.05059\n",
      "Time:                        17:55:35   Log-Likelihood:                -1710.0\n",
      "converged:                       True   LL-Null:                       -1801.1\n",
      "                                        LLR p-value:                 1.150e-36\n",
      "==========================================================================================\n",
      "     chosen_mode3=uber       coef    std err          z      P>|z|      [95.0% Conf. Int.]\n",
      "------------------------------------------------------------------------------------------\n",
      "n_transit                 -0.5462      0.059     -9.245      0.000        -0.662    -0.430\n",
      "walkwait                   0.0007      0.000      6.549      0.000         0.000     0.001\n",
      "uber_basic_time           -0.0134      0.010     -1.319      0.187        -0.033     0.007\n",
      "uber_basic_cost            0.0195      0.011      1.707      0.088        -0.003     0.042\n",
      "start_or_end_near_home     0.2537      0.109      2.318      0.020         0.039     0.468\n",
      "start_or_end_near_work     0.6007      0.107      5.594      0.000         0.390     0.811\n",
      "const                     -2.9659      0.197    -15.030      0.000        -3.353    -2.579\n",
      "==========================================================================================\n"
     ]
    }
   ],
   "source": [
    "# --------------------------\n",
    "# - SERIES 3               - \n",
    "# --------------------------\n",
    "# drop transit_time, transit_cost\n",
    "exog_cols = ['n_transit','walkwait',\n",
    "             'uber_basic_time','uber_basic_cost',\n",
    "             'start_or_end_near_home','start_or_end_near_work']\n",
    "defn = subset2.dropna(subset=exog_cols)\n",
    "y = defn['chosen_mode3']\n",
    "x = defn.loc[:,exog_cols]\n",
    "x['const'] = 1\n",
    "\n",
    "s2_mdl = st.MNLogit(y, x)\n",
    "s2_fit = s2_mdl.fit(maxiter=100)\n",
    "print s2_fit.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Optimization terminated successfully.\n",
      "         Current function value: 0.106765\n",
      "         Iterations 8\n",
      "                          MNLogit Regression Results                          \n",
      "==============================================================================\n",
      "Dep. Variable:           chosen_mode3   No. Observations:                16031\n",
      "Model:                        MNLogit   Df Residuals:                    16026\n",
      "Method:                           MLE   Df Model:                            4\n",
      "Date:                Wed, 20 Jul 2016   Pseudo R-squ.:                 0.04976\n",
      "Time:                        17:55:35   Log-Likelihood:                -1711.6\n",
      "converged:                       True   LL-Null:                       -1801.2\n",
      "                                        LLR p-value:                 1.081e-37\n",
      "==========================================================================================\n",
      "     chosen_mode3=uber       coef    std err          z      P>|z|      [95.0% Conf. Int.]\n",
      "------------------------------------------------------------------------------------------\n",
      "n_transit                 -0.5659      0.058     -9.758      0.000        -0.680    -0.452\n",
      "walkwait                   0.0007      0.000      6.769      0.000         0.001     0.001\n",
      "start_or_end_near_home     0.2435      0.109      2.229      0.026         0.029     0.458\n",
      "start_or_end_near_work     0.5787      0.105      5.518      0.000         0.373     0.784\n",
      "const                     -2.8788      0.177    -16.262      0.000        -3.226    -2.532\n",
      "==========================================================================================\n"
     ]
    }
   ],
   "source": [
    "# --------------------------\n",
    "# - SERIES 3               - \n",
    "# --------------------------\n",
    "# drop uber_basic_time, uber_basic_cost\n",
    "exog_cols = ['n_transit','walkwait',\n",
    "             'start_or_end_near_home','start_or_end_near_work']\n",
    "defn = subset2.dropna(subset=exog_cols)\n",
    "y = defn['chosen_mode3']\n",
    "x = defn.loc[:,exog_cols]\n",
    "x['const'] = 1\n",
    "\n",
    "s2_mdl = st.MNLogit(y, x)\n",
    "s2_fit = s2_mdl.fit(maxiter=100)\n",
    "print s2_fit.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {
    "collapsed": false,
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Optimization terminated successfully.\n",
      "         Current function value: 0.106574\n",
      "         Iterations 8\n",
      "                          MNLogit Regression Results                          \n",
      "==============================================================================\n",
      "Dep. Variable:           chosen_mode3   No. Observations:                15921\n",
      "Model:                        MNLogit   Df Residuals:                    15910\n",
      "Method:                           MLE   Df Model:                           10\n",
      "Date:                Wed, 20 Jul 2016   Pseudo R-squ.:                 0.05267\n",
      "Time:                        17:55:35   Log-Likelihood:                -1696.8\n",
      "converged:                       True   LL-Null:                       -1791.1\n",
      "                                        LLR p-value:                 3.696e-35\n",
      "==========================================================================================\n",
      "     chosen_mode3=uber       coef    std err          z      P>|z|      [95.0% Conf. Int.]\n",
      "------------------------------------------------------------------------------------------\n",
      "n_transit                 -0.5553      0.060     -9.289      0.000        -0.673    -0.438\n",
      "n_arr                     -0.0039      0.142     -0.028      0.978        -0.281     0.273\n",
      "ln_walkwait                0.1503      0.060      2.504      0.012         0.033     0.268\n",
      "ln_headway                -0.0011      0.104     -0.011      0.992        -0.204     0.202\n",
      "ln_transit_time           -0.2633      0.147     -1.792      0.073        -0.551     0.025\n",
      "ln_uber_basic_time         0.0719      0.175      0.412      0.680        -0.270     0.414\n",
      "transit_cost               0.0218      0.044      0.494      0.621        -0.065     0.108\n",
      "uber_basic_cost            0.0087      0.010      0.855      0.392        -0.011     0.029\n",
      "start_or_end_near_home     0.2888      0.112      2.575      0.010         0.069     0.509\n",
      "start_or_end_near_work     0.6051      0.109      5.541      0.000         0.391     0.819\n",
      "const                     -2.7952      0.756     -3.699      0.000        -4.276    -1.314\n",
      "==========================================================================================\n"
     ]
    }
   ],
   "source": [
    "# --------------------------\n",
    "# - SERIES 4               - \n",
    "# --------------------------\n",
    "# Start with SERIES 3\n",
    "# log the walkwait, headway, transit_time, uber_basic_time\n",
    "exog_cols = ['n_transit','n_arr','ln_walkwait','ln_headway','ln_transit_time',\n",
    "             'ln_uber_basic_time','transit_cost','uber_basic_cost',\n",
    "             'start_or_end_near_home','start_or_end_near_work']\n",
    "defn = subset2.dropna(subset=exog_cols)\n",
    "y = defn['chosen_mode3']\n",
    "x = defn.loc[:,exog_cols]\n",
    "x['const'] = 1\n",
    "\n",
    "s2_mdl = st.MNLogit(y, x)\n",
    "s2_fit = s2_mdl.fit(maxiter=100)\n",
    "print s2_fit.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Optimization terminated successfully.\n",
      "         Current function value: 0.106509\n",
      "         Iterations 8\n",
      "                          MNLogit Regression Results                          \n",
      "==============================================================================\n",
      "Dep. Variable:           chosen_mode3   No. Observations:                15977\n",
      "Model:                        MNLogit   Df Residuals:                    15968\n",
      "Method:                           MLE   Df Model:                            8\n",
      "Date:                Wed, 20 Jul 2016   Pseudo R-squ.:                 0.05259\n",
      "Time:                        17:55:36   Log-Likelihood:                -1701.7\n",
      "converged:                       True   LL-Null:                       -1796.2\n",
      "                                        LLR p-value:                 1.371e-36\n",
      "==========================================================================================\n",
      "     chosen_mode3=uber       coef    std err          z      P>|z|      [95.0% Conf. Int.]\n",
      "------------------------------------------------------------------------------------------\n",
      "n_transit                 -0.5534      0.060     -9.275      0.000        -0.670    -0.436\n",
      "ln_walkwait                0.1524      0.022      6.802      0.000         0.108     0.196\n",
      "ln_transit_time           -0.2682      0.146     -1.832      0.067        -0.555     0.019\n",
      "ln_uber_basic_time         0.0747      0.174      0.430      0.667        -0.266     0.415\n",
      "transit_cost               0.0230      0.044      0.524      0.600        -0.063     0.109\n",
      "uber_basic_cost            0.0085      0.010      0.842      0.400        -0.011     0.028\n",
      "start_or_end_near_home     0.2905      0.112      2.593      0.010         0.071     0.510\n",
      "start_or_end_near_work     0.6096      0.109      5.590      0.000         0.396     0.823\n",
      "const                     -2.8165      0.380     -7.421      0.000        -3.560    -2.073\n",
      "==========================================================================================\n"
     ]
    }
   ],
   "source": [
    "# --------------------------\n",
    "# - SERIES 4               - \n",
    "# --------------------------\n",
    "# drop n_arr, headway\n",
    "exog_cols = ['n_transit','ln_walkwait','ln_transit_time',\n",
    "             'ln_uber_basic_time','transit_cost','uber_basic_cost',\n",
    "             'start_or_end_near_home','start_or_end_near_work']\n",
    "defn = subset2.dropna(subset=exog_cols)\n",
    "y = defn['chosen_mode3']\n",
    "x = defn.loc[:,exog_cols]\n",
    "x['const'] = 1\n",
    "\n",
    "s2_mdl = st.MNLogit(y, x)\n",
    "s2_fit = s2_mdl.fit(maxiter=100)\n",
    "print s2_fit.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Optimization terminated successfully.\n",
      "         Current function value: 0.106521\n",
      "         Iterations 8\n",
      "                          MNLogit Regression Results                          \n",
      "==============================================================================\n",
      "Dep. Variable:           chosen_mode3   No. Observations:                15977\n",
      "Model:                        MNLogit   Df Residuals:                    15970\n",
      "Method:                           MLE   Df Model:                            6\n",
      "Date:                Wed, 20 Jul 2016   Pseudo R-squ.:                 0.05249\n",
      "Time:                        17:55:36   Log-Likelihood:                -1701.9\n",
      "converged:                       True   LL-Null:                       -1796.2\n",
      "                                        LLR p-value:                 5.147e-38\n",
      "==========================================================================================\n",
      "     chosen_mode3=uber       coef    std err          z      P>|z|      [95.0% Conf. Int.]\n",
      "------------------------------------------------------------------------------------------\n",
      "n_transit                 -0.5496      0.059     -9.278      0.000        -0.666    -0.433\n",
      "ln_walkwait                0.1525      0.022      6.827      0.000         0.109     0.196\n",
      "ln_transit_time           -0.2313      0.119     -1.943      0.052        -0.465     0.002\n",
      "uber_basic_cost            0.0132      0.006      2.067      0.039         0.001     0.026\n",
      "start_or_end_near_home     0.2861      0.112      2.560      0.010         0.067     0.505\n",
      "start_or_end_near_work     0.6141      0.107      5.765      0.000         0.405     0.823\n",
      "const                     -2.7354      0.348     -7.864      0.000        -3.417    -2.054\n",
      "==========================================================================================\n"
     ]
    }
   ],
   "source": [
    "# --------------------------\n",
    "# - SERIES 4               - \n",
    "# --------------------------\n",
    "# drop uber_basic_time, transit_cost\n",
    "exog_cols = ['n_transit','ln_walkwait','ln_transit_time',\n",
    "             'uber_basic_cost',\n",
    "             'start_or_end_near_home','start_or_end_near_work']\n",
    "defn = subset2.dropna(subset=exog_cols)\n",
    "y = defn['chosen_mode3']\n",
    "x = defn.loc[:,exog_cols]\n",
    "x['const'] = 1\n",
    "\n",
    "s2_mdl = st.MNLogit(y, x)\n",
    "s2_fit = s2_mdl.fit(maxiter=100)\n",
    "print s2_fit.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Optimization terminated successfully.\n",
      "         Current function value: 0.106640\n",
      "         Iterations 8\n",
      "                          MNLogit Regression Results                          \n",
      "==============================================================================\n",
      "Dep. Variable:           chosen_mode3   No. Observations:                15982\n",
      "Model:                        MNLogit   Df Residuals:                    15977\n",
      "Method:                           MLE   Df Model:                            4\n",
      "Date:                Wed, 20 Jul 2016   Pseudo R-squ.:                 0.05120\n",
      "Time:                        17:55:36   Log-Likelihood:                -1704.3\n",
      "converged:                       True   LL-Null:                       -1796.3\n",
      "                                        LLR p-value:                 1.064e-38\n",
      "==========================================================================================\n",
      "     chosen_mode3=uber       coef    std err          z      P>|z|      [95.0% Conf. Int.]\n",
      "------------------------------------------------------------------------------------------\n",
      "n_transit                 -0.5744      0.058     -9.908      0.000        -0.688    -0.461\n",
      "ln_walkwait                0.1536      0.022      6.922      0.000         0.110     0.197\n",
      "start_or_end_near_home     0.2425      0.110      2.214      0.027         0.028     0.457\n",
      "start_or_end_near_work     0.5916      0.105      5.633      0.000         0.386     0.797\n",
      "const                     -3.1993      0.196    -16.310      0.000        -3.584    -2.815\n",
      "==========================================================================================\n"
     ]
    }
   ],
   "source": [
    "# --------------------------\n",
    "# - SERIES 4               - \n",
    "# --------------------------\n",
    "# replace ln_walkwait with walkwait\n",
    "exog_cols = ['n_transit','ln_walkwait',\n",
    "             'start_or_end_near_home','start_or_end_near_work']\n",
    "defn = subset2.dropna(subset=exog_cols)\n",
    "y = defn['chosen_mode3']\n",
    "x = defn.loc[:,exog_cols]\n",
    "x['const'] = 1\n",
    "\n",
    "s2_mdl = st.MNLogit(y, x)\n",
    "s2_fit = s2_mdl.fit(maxiter=100)\n",
    "print s2_fit.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Optimization terminated successfully.\n",
      "         Current function value: 0.106372\n",
      "         Iterations 8\n",
      "                          MNLogit Regression Results                          \n",
      "==============================================================================\n",
      "Dep. Variable:           chosen_mode3   No. Observations:                15921\n",
      "Model:                        MNLogit   Df Residuals:                    15909\n",
      "Method:                           MLE   Df Model:                           11\n",
      "Date:                Wed, 20 Jul 2016   Pseudo R-squ.:                 0.05447\n",
      "Time:                        17:55:36   Log-Likelihood:                -1693.5\n",
      "converged:                       True   LL-Null:                       -1791.1\n",
      "                                        LLR p-value:                 7.661e-36\n",
      "==========================================================================================\n",
      "     chosen_mode3=uber       coef    std err          z      P>|z|      [95.0% Conf. Int.]\n",
      "------------------------------------------------------------------------------------------\n",
      "n_transit                 -0.5496      0.060     -9.197      0.000        -0.667    -0.432\n",
      "n_arr                     -0.0091      0.141     -0.065      0.948        -0.285     0.267\n",
      "ln_walkwait                0.1567      0.060      2.606      0.009         0.039     0.275\n",
      "ln_headway                -0.0260      0.104     -0.251      0.802        -0.229     0.177\n",
      "ln_transit_time           -0.3011      0.147     -2.047      0.041        -0.589    -0.013\n",
      "ln_uber_basic_time         0.1274      0.177      0.722      0.470        -0.219     0.473\n",
      "transit_cost               0.0229      0.044      0.519      0.604        -0.064     0.109\n",
      "uber_basic_cost            0.0078      0.010      0.760      0.447        -0.012     0.028\n",
      "start_or_end_near_home     0.2739      0.112      2.436      0.015         0.054     0.494\n",
      "start_or_end_near_work     0.6169      0.109      5.641      0.000         0.403     0.831\n",
      "late_night                 0.5289      0.197      2.689      0.007         0.143     0.914\n",
      "const                     -2.7983      0.752     -3.720      0.000        -4.273    -1.324\n",
      "==========================================================================================\n"
     ]
    }
   ],
   "source": [
    "# --------------------------\n",
    "# - SERIES 5               - \n",
    "# --------------------------\n",
    "# Start with SERIES 4\n",
    "# add late_night\n",
    "exog_cols = ['n_transit','n_arr','ln_walkwait','ln_headway','ln_transit_time',\n",
    "             'ln_uber_basic_time','transit_cost','uber_basic_cost',\n",
    "             'start_or_end_near_home','start_or_end_near_work','late_night']\n",
    "defn = subset2.dropna(subset=exog_cols)\n",
    "y = defn['chosen_mode3']\n",
    "x = defn.loc[:,exog_cols]\n",
    "x['const'] = 1\n",
    "\n",
    "s2_mdl = st.MNLogit(y, x)\n",
    "s2_fit = s2_mdl.fit(maxiter=100)\n",
    "print s2_fit.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Optimization terminated successfully.\n",
      "         Current function value: 0.106316\n",
      "         Iterations 8\n",
      "                          MNLogit Regression Results                          \n",
      "==============================================================================\n",
      "Dep. Variable:           chosen_mode3   No. Observations:                15977\n",
      "Model:                        MNLogit   Df Residuals:                    15967\n",
      "Method:                           MLE   Df Model:                            9\n",
      "Date:                Wed, 20 Jul 2016   Pseudo R-squ.:                 0.05431\n",
      "Time:                        17:55:36   Log-Likelihood:                -1698.6\n",
      "converged:                       True   LL-Null:                       -1796.2\n",
      "                                        LLR p-value:                 3.537e-37\n",
      "==========================================================================================\n",
      "     chosen_mode3=uber       coef    std err          z      P>|z|      [95.0% Conf. Int.]\n",
      "------------------------------------------------------------------------------------------\n",
      "n_transit                 -0.5482      0.060     -9.194      0.000        -0.665    -0.431\n",
      "ln_walkwait                0.1511      0.022      6.724      0.000         0.107     0.195\n",
      "ln_transit_time           -0.3047      0.146     -2.081      0.037        -0.592    -0.018\n",
      "ln_uber_basic_time         0.1327      0.176      0.755      0.450        -0.212     0.477\n",
      "transit_cost               0.0234      0.044      0.533      0.594        -0.063     0.109\n",
      "uber_basic_cost            0.0075      0.010      0.734      0.463        -0.013     0.027\n",
      "start_or_end_near_home     0.2757      0.112      2.455      0.014         0.056     0.496\n",
      "start_or_end_near_work     0.6219      0.109      5.693      0.000         0.408     0.836\n",
      "late_night                 0.5124      0.195      2.631      0.009         0.131     0.894\n",
      "const                     -2.8845      0.382     -7.547      0.000        -3.634    -2.135\n",
      "==========================================================================================\n"
     ]
    }
   ],
   "source": [
    "# --------------------------\n",
    "# - SERIES 5               - \n",
    "# --------------------------\n",
    "# drop n_arr, headeay\n",
    "exog_cols = ['n_transit','ln_walkwait','ln_transit_time',\n",
    "             'ln_uber_basic_time','transit_cost','uber_basic_cost',\n",
    "             'start_or_end_near_home','start_or_end_near_work','late_night']\n",
    "defn = subset2.dropna(subset=exog_cols)\n",
    "y = defn['chosen_mode3']\n",
    "x = defn.loc[:,exog_cols]\n",
    "x['const'] = 1\n",
    "\n",
    "s2_mdl = st.MNLogit(y, x)\n",
    "s2_fit = s2_mdl.fit(maxiter=100)\n",
    "print s2_fit.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Optimization terminated successfully.\n",
      "         Current function value: 0.106338\n",
      "         Iterations 8\n",
      "                          MNLogit Regression Results                          \n",
      "==============================================================================\n",
      "Dep. Variable:           chosen_mode3   No. Observations:                15977\n",
      "Model:                        MNLogit   Df Residuals:                    15969\n",
      "Method:                           MLE   Df Model:                            7\n",
      "Date:                Wed, 20 Jul 2016   Pseudo R-squ.:                 0.05411\n",
      "Time:                        17:55:37   Log-Likelihood:                -1699.0\n",
      "converged:                       True   LL-Null:                       -1796.2\n",
      "                                        LLR p-value:                 1.761e-38\n",
      "==========================================================================================\n",
      "     chosen_mode3=uber       coef    std err          z      P>|z|      [95.0% Conf. Int.]\n",
      "------------------------------------------------------------------------------------------\n",
      "n_transit                 -0.5426      0.059     -9.166      0.000        -0.659    -0.427\n",
      "ln_walkwait                0.1508      0.022      6.727      0.000         0.107     0.195\n",
      "ln_transit_time           -0.2398      0.119     -2.018      0.044        -0.473    -0.007\n",
      "uber_basic_cost            0.0137      0.006      2.132      0.033         0.001     0.026\n",
      "start_or_end_near_home     0.2706      0.112      2.416      0.016         0.051     0.490\n",
      "start_or_end_near_work     0.6334      0.107      5.923      0.000         0.424     0.843\n",
      "late_night                 0.4927      0.192      2.561      0.010         0.116     0.870\n",
      "const                     -2.7569      0.347     -7.936      0.000        -3.438    -2.076\n",
      "==========================================================================================\n"
     ]
    }
   ],
   "source": [
    "# --------------------------\n",
    "# - SERIES 5               - \n",
    "# --------------------------\n",
    "# drop uber_basic_time, transit_cost\n",
    "exog_cols = ['n_transit','ln_walkwait','ln_transit_time',\n",
    "             'uber_basic_cost',\n",
    "             'start_or_end_near_home','start_or_end_near_work','late_night']\n",
    "defn = subset2.dropna(subset=exog_cols)\n",
    "y = defn['chosen_mode3']\n",
    "x = defn.loc[:,exog_cols]\n",
    "x['const'] = 1\n",
    "\n",
    "s2_mdl = st.MNLogit(y, x)\n",
    "s2_fit = s2_mdl.fit(maxiter=100)\n",
    "print s2_fit.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Optimization terminated successfully.\n",
      "         Current function value: 0.106467\n",
      "         Iterations 8\n",
      "                          MNLogit Regression Results                          \n",
      "==============================================================================\n",
      "Dep. Variable:           chosen_mode3   No. Observations:                15982\n",
      "Model:                        MNLogit   Df Residuals:                    15976\n",
      "Method:                           MLE   Df Model:                            5\n",
      "Date:                Wed, 20 Jul 2016   Pseudo R-squ.:                 0.05274\n",
      "Time:                        17:55:37   Log-Likelihood:                -1701.5\n",
      "converged:                       True   LL-Null:                       -1796.3\n",
      "                                        LLR p-value:                 5.073e-39\n",
      "==========================================================================================\n",
      "     chosen_mode3=uber       coef    std err          z      P>|z|      [95.0% Conf. Int.]\n",
      "------------------------------------------------------------------------------------------\n",
      "n_transit                 -0.5681      0.058     -9.803      0.000        -0.682    -0.455\n",
      "ln_walkwait                0.1519      0.022      6.820      0.000         0.108     0.196\n",
      "start_or_end_near_home     0.2263      0.110      2.061      0.039         0.011     0.442\n",
      "start_or_end_near_work     0.6092      0.105      5.778      0.000         0.403     0.816\n",
      "late_night                 0.4785      0.192      2.489      0.013         0.102     0.855\n",
      "const                     -3.2379      0.197    -16.423      0.000        -3.624    -2.851\n",
      "==========================================================================================\n"
     ]
    }
   ],
   "source": [
    "# --------------------------\n",
    "# - SERIES 5               - \n",
    "# --------------------------\n",
    "# drop transit_time, uber_basic_cost\n",
    "exog_cols = ['n_transit','ln_walkwait',\n",
    "             'start_or_end_near_home','start_or_end_near_work','late_night']\n",
    "defn = subset2.dropna(subset=exog_cols)\n",
    "y = defn['chosen_mode3']\n",
    "x = defn.loc[:,exog_cols]\n",
    "x['const'] = 1\n",
    "\n",
    "s2_mdl = st.MNLogit(y, x)\n",
    "s2_fit = s2_mdl.fit(maxiter=100)\n",
    "print s2_fit.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Optimization terminated successfully.\n",
      "         Current function value: 0.106499\n",
      "         Iterations 8\n",
      "                          MNLogit Regression Results                          \n",
      "==============================================================================\n",
      "Dep. Variable:           chosen_mode3   No. Observations:                16026\n",
      "Model:                        MNLogit   Df Residuals:                    16014\n",
      "Method:                           MLE   Df Model:                           11\n",
      "Date:                Wed, 20 Jul 2016   Pseudo R-squ.:                 0.05236\n",
      "Time:                        17:55:37   Log-Likelihood:                -1706.8\n",
      "converged:                       True   LL-Null:                       -1801.1\n",
      "                                        LLR p-value:                 1.708e-34\n",
      "==========================================================================================\n",
      "     chosen_mode3=uber       coef    std err          z      P>|z|      [95.0% Conf. Int.]\n",
      "------------------------------------------------------------------------------------------\n",
      "n_transit                 -0.5396      0.059     -9.088      0.000        -0.656    -0.423\n",
      "n_arr                     -0.0127      0.218     -0.058      0.954        -0.441     0.415\n",
      "walkwait                   0.0007      0.000      2.129      0.033      5.23e-05     0.001\n",
      "headway                    0.0002      0.005      0.031      0.975        -0.010     0.010\n",
      "transit_time              -0.0038      0.005     -0.807      0.420        -0.013     0.005\n",
      "uber_basic_time           -0.0091      0.011     -0.854      0.393        -0.030     0.012\n",
      "transit_cost               0.0177      0.044      0.401      0.688        -0.069     0.104\n",
      "uber_basic_cost            0.0182      0.015      1.179      0.239        -0.012     0.049\n",
      "start_or_end_near_home     0.2507      0.112      2.246      0.025         0.032     0.469\n",
      "start_or_end_near_work     0.6192      0.108      5.729      0.000         0.407     0.831\n",
      "late_night                 0.4943      0.197      2.508      0.012         0.108     0.881\n",
      "const                     -2.9948      0.706     -4.243      0.000        -4.378    -1.612\n",
      "==========================================================================================\n"
     ]
    }
   ],
   "source": [
    "# --------------------------\n",
    "# - SERIES 6               - \n",
    "# --------------------------\n",
    "# start with SERIES 3, add late_night\n",
    "exog_cols = ['n_transit','n_arr','walkwait','headway','transit_time',\n",
    "             'uber_basic_time','transit_cost','uber_basic_cost',\n",
    "             'start_or_end_near_home','start_or_end_near_work','late_night']\n",
    "defn = subset2.dropna(subset=exog_cols)\n",
    "y = defn['chosen_mode3']\n",
    "x = defn.loc[:,exog_cols]\n",
    "x['const'] = 1\n",
    "\n",
    "s2_mdl = st.MNLogit(y, x)\n",
    "s2_fit = s2_mdl.fit(maxiter=100)\n",
    "print s2_fit.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Optimization terminated successfully.\n",
      "         Current function value: 0.106500\n",
      "         Iterations 8\n",
      "                          MNLogit Regression Results                          \n",
      "==============================================================================\n",
      "Dep. Variable:           chosen_mode3   No. Observations:                16026\n",
      "Model:                        MNLogit   Df Residuals:                    16016\n",
      "Method:                           MLE   Df Model:                            9\n",
      "Date:                Wed, 20 Jul 2016   Pseudo R-squ.:                 0.05235\n",
      "Time:                        17:55:37   Log-Likelihood:                -1706.8\n",
      "converged:                       True   LL-Null:                       -1801.1\n",
      "                                        LLR p-value:                 8.197e-36\n",
      "==========================================================================================\n",
      "     chosen_mode3=uber       coef    std err          z      P>|z|      [95.0% Conf. Int.]\n",
      "------------------------------------------------------------------------------------------\n",
      "n_transit                 -0.5394      0.059     -9.088      0.000        -0.656    -0.423\n",
      "walkwait                   0.0007      0.000      6.515      0.000         0.000     0.001\n",
      "transit_time              -0.0038      0.005     -0.812      0.417        -0.013     0.005\n",
      "uber_basic_time           -0.0092      0.011     -0.872      0.383        -0.030     0.012\n",
      "transit_cost               0.0182      0.044      0.413      0.680        -0.068     0.104\n",
      "uber_basic_cost            0.0184      0.015      1.191      0.234        -0.012     0.049\n",
      "start_or_end_near_home     0.2512      0.112      2.252      0.024         0.033     0.470\n",
      "start_or_end_near_work     0.6190      0.108      5.728      0.000         0.407     0.831\n",
      "late_night                 0.4993      0.195      2.557      0.011         0.117     0.882\n",
      "const                     -3.0263      0.206    -14.723      0.000        -3.429    -2.623\n",
      "==========================================================================================\n"
     ]
    }
   ],
   "source": [
    "# --------------------------\n",
    "# - SERIES 6               - \n",
    "# --------------------------\n",
    "# drop n_arr, headway\n",
    "exog_cols = ['n_transit','walkwait','transit_time',\n",
    "             'uber_basic_time','transit_cost','uber_basic_cost',\n",
    "             'start_or_end_near_home','start_or_end_near_work','late_night']\n",
    "defn = subset2.dropna(subset=exog_cols)\n",
    "y = defn['chosen_mode3']\n",
    "x = defn.loc[:,exog_cols]\n",
    "x['const'] = 1\n",
    "\n",
    "s2_mdl = st.MNLogit(y, x)\n",
    "s2_fit = s2_mdl.fit(maxiter=100)\n",
    "print s2_fit.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Optimization terminated successfully.\n",
      "         Current function value: 0.106524\n",
      "         Iterations 8\n",
      "                          MNLogit Regression Results                          \n",
      "==============================================================================\n",
      "Dep. Variable:           chosen_mode3   No. Observations:                16026\n",
      "Model:                        MNLogit   Df Residuals:                    16018\n",
      "Method:                           MLE   Df Model:                            7\n",
      "Date:                Wed, 20 Jul 2016   Pseudo R-squ.:                 0.05214\n",
      "Time:                        17:55:37   Log-Likelihood:                -1707.1\n",
      "converged:                       True   LL-Null:                       -1801.1\n",
      "                                        LLR p-value:                 4.330e-37\n",
      "==========================================================================================\n",
      "     chosen_mode3=uber       coef    std err          z      P>|z|      [95.0% Conf. Int.]\n",
      "------------------------------------------------------------------------------------------\n",
      "n_transit                 -0.5421      0.059     -9.175      0.000        -0.658    -0.426\n",
      "walkwait                   0.0007      0.000      6.540      0.000         0.000     0.001\n",
      "uber_basic_time           -0.0098      0.010     -0.947      0.343        -0.030     0.010\n",
      "uber_basic_cost            0.0157      0.012      1.355      0.175        -0.007     0.039\n",
      "start_or_end_near_home     0.2348      0.110      2.138      0.033         0.020     0.450\n",
      "start_or_end_near_work     0.6108      0.108      5.680      0.000         0.400     0.822\n",
      "late_night                 0.4877      0.195      2.502      0.012         0.106     0.870\n",
      "const                     -3.0141      0.199    -15.142      0.000        -3.404    -2.624\n",
      "==========================================================================================\n"
     ]
    }
   ],
   "source": [
    "# --------------------------\n",
    "# - SERIES 6               - \n",
    "# --------------------------\n",
    "# drop transit_time, transit_cost\n",
    "exog_cols = ['n_transit','walkwait',\n",
    "             'uber_basic_time','uber_basic_cost',\n",
    "             'start_or_end_near_home','start_or_end_near_work','late_night']\n",
    "defn = subset2.dropna(subset=exog_cols)\n",
    "y = defn['chosen_mode3']\n",
    "x = defn.loc[:,exog_cols]\n",
    "x['const'] = 1\n",
    "\n",
    "s2_mdl = st.MNLogit(y, x)\n",
    "s2_fit = s2_mdl.fit(maxiter=100)\n",
    "print s2_fit.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 88,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Optimization terminated successfully.\n",
      "         Current function value: 0.106564\n",
      "         Iterations 8\n",
      "                          MNLogit Regression Results                          \n",
      "==============================================================================\n",
      "Dep. Variable:           chosen_mode3   No. Observations:                16031\n",
      "Model:                        MNLogit   Df Residuals:                    16025\n",
      "Method:                           MLE   Df Model:                            5\n",
      "Date:                Wed, 20 Jul 2016   Pseudo R-squ.:                 0.05155\n",
      "Time:                        18:05:24   Log-Likelihood:                -1708.3\n",
      "converged:                       True   LL-Null:                       -1801.2\n",
      "                                        LLR p-value:                 3.235e-38\n",
      "==========================================================================================\n",
      "     chosen_mode3=uber       coef    std err          z      P>|z|      [95.0% Conf. Int.]\n",
      "------------------------------------------------------------------------------------------\n",
      "n_transit                 -0.5596      0.058     -9.654      0.000        -0.673    -0.446\n",
      "walkwait                   0.0007      0.000      6.733      0.000         0.001     0.001\n",
      "start_or_end_near_home     0.2262      0.110      2.065      0.039         0.011     0.441\n",
      "start_or_end_near_work     0.5979      0.105      5.678      0.000         0.392     0.804\n",
      "late_night                 0.5189      0.192      2.703      0.007         0.143     0.895\n",
      "const                     -2.9255      0.178    -16.419      0.000        -3.275    -2.576\n",
      "==========================================================================================\n"
     ]
    }
   ],
   "source": [
    "# --------------------------\n",
    "# - SERIES 6               - \n",
    "# --------------------------\n",
    "# drop uber_basic_time, uber_basic_cost\n",
    "exog_cols = ['n_transit','walkwait',\n",
    "             'start_or_end_near_home','start_or_end_near_work','late_night']\n",
    "defn = subset2.dropna(subset=exog_cols)\n",
    "y = defn['chosen_mode3']\n",
    "x = defn.loc[:,exog_cols]\n",
    "x['const'] = 1\n",
    "\n",
    "s2_mdl = st.MNLogit(y, x)\n",
    "s2_fit = s2_mdl.fit(maxiter=100)\n",
    "print s2_fit.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "metadata": {
    "collapsed": false,
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Optimization terminated successfully.\n",
      "         Current function value: 0.105683\n",
      "         Iterations 8\n",
      "Optimization terminated successfully.\n",
      "         Current function value: 0.106315\n",
      "         Iterations 8\n",
      "Optimization terminated successfully.\n",
      "         Current function value: 0.106765\n",
      "         Iterations 8\n",
      "Optimization terminated successfully.\n",
      "         Current function value: 0.106640\n",
      "         Iterations 8\n",
      "Optimization terminated successfully.\n",
      "         Current function value: 0.106467\n",
      "         Iterations 8\n",
      "Optimization terminated successfully.\n",
      "         Current function value: 0.106564\n",
      "         Iterations 8\n",
      "SERIES 1 FINAL MODEL\n",
      "                          MNLogit Regression Results                          \n",
      "==============================================================================\n",
      "Dep. Variable:           chosen_mode3   No. Observations:                19646\n",
      "Model:                        MNLogit   Df Residuals:                    19642\n",
      "Method:                           MLE   Df Model:                            3\n",
      "Date:                Wed, 20 Jul 2016   Pseudo R-squ.:                 0.04338\n",
      "Time:                        17:55:38   Log-Likelihood:                -2076.2\n",
      "converged:                       True   LL-Null:                       -2170.4\n",
      "                                        LLR p-value:                 1.420e-40\n",
      "=====================================================================================\n",
      "chosen_mode3=uber       coef    std err          z      P>|z|      [95.0% Conf. Int.]\n",
      "-------------------------------------------------------------------------------------\n",
      "n_transit            -0.5690      0.053    -10.787      0.000        -0.672    -0.466\n",
      "wait                  0.0006   9.89e-05      6.234      0.000         0.000     0.001\n",
      "transit_cost          0.0610      0.020      3.053      0.002         0.022     0.100\n",
      "const                -2.6575      0.158    -16.807      0.000        -2.967    -2.348\n",
      "=====================================================================================\n",
      "SERIES 2 FINAL MODEL\n",
      "                          MNLogit Regression Results                          \n",
      "==============================================================================\n",
      "Dep. Variable:           chosen_mode3   No. Observations:                16259\n",
      "Model:                        MNLogit   Df Residuals:                    16254\n",
      "Method:                           MLE   Df Model:                            4\n",
      "Date:                Wed, 20 Jul 2016   Pseudo R-squ.:                 0.04909\n",
      "Time:                        17:55:38   Log-Likelihood:                -1728.6\n",
      "converged:                       True   LL-Null:                       -1817.8\n",
      "                                        LLR p-value:                 1.592e-37\n",
      "======================================================================================\n",
      " chosen_mode3=uber       coef    std err          z      P>|z|      [95.0% Conf. Int.]\n",
      "--------------------------------------------------------------------------------------\n",
      "n_transit             -0.5597      0.058     -9.692      0.000        -0.673    -0.446\n",
      "wait                   0.0007      0.000      6.757      0.000         0.001     0.001\n",
      "start_is_near_work     0.3469      0.150      2.319      0.020         0.054     0.640\n",
      "end_is_near_work       0.6833      0.116      5.875      0.000         0.455     0.911\n",
      "const                 -2.7394      0.162    -16.896      0.000        -3.057    -2.422\n",
      "======================================================================================\n",
      "SERIES 3 FINAL MODEL\n",
      "                          MNLogit Regression Results                          \n",
      "==============================================================================\n",
      "Dep. Variable:           chosen_mode3   No. Observations:                16031\n",
      "Model:                        MNLogit   Df Residuals:                    16026\n",
      "Method:                           MLE   Df Model:                            4\n",
      "Date:                Wed, 20 Jul 2016   Pseudo R-squ.:                 0.04976\n",
      "Time:                        17:55:38   Log-Likelihood:                -1711.6\n",
      "converged:                       True   LL-Null:                       -1801.2\n",
      "                                        LLR p-value:                 1.081e-37\n",
      "==========================================================================================\n",
      "     chosen_mode3=uber       coef    std err          z      P>|z|      [95.0% Conf. Int.]\n",
      "------------------------------------------------------------------------------------------\n",
      "n_transit                 -0.5659      0.058     -9.758      0.000        -0.680    -0.452\n",
      "walkwait                   0.0007      0.000      6.769      0.000         0.001     0.001\n",
      "start_or_end_near_home     0.2435      0.109      2.229      0.026         0.029     0.458\n",
      "start_or_end_near_work     0.5787      0.105      5.518      0.000         0.373     0.784\n",
      "const                     -2.8788      0.177    -16.262      0.000        -3.226    -2.532\n",
      "==========================================================================================\n",
      "SERIES 4 FINAL MODEL\n",
      "                          MNLogit Regression Results                          \n",
      "==============================================================================\n",
      "Dep. Variable:           chosen_mode3   No. Observations:                15982\n",
      "Model:                        MNLogit   Df Residuals:                    15977\n",
      "Method:                           MLE   Df Model:                            4\n",
      "Date:                Wed, 20 Jul 2016   Pseudo R-squ.:                 0.05120\n",
      "Time:                        17:55:38   Log-Likelihood:                -1704.3\n",
      "converged:                       True   LL-Null:                       -1796.3\n",
      "                                        LLR p-value:                 1.064e-38\n",
      "==========================================================================================\n",
      "     chosen_mode3=uber       coef    std err          z      P>|z|      [95.0% Conf. Int.]\n",
      "------------------------------------------------------------------------------------------\n",
      "n_transit                 -0.5744      0.058     -9.908      0.000        -0.688    -0.461\n",
      "ln_walkwait                0.1536      0.022      6.922      0.000         0.110     0.197\n",
      "start_or_end_near_home     0.2425      0.110      2.214      0.027         0.028     0.457\n",
      "start_or_end_near_work     0.5916      0.105      5.633      0.000         0.386     0.797\n",
      "const                     -3.1993      0.196    -16.310      0.000        -3.584    -2.815\n",
      "==========================================================================================\n",
      "SERIES 5 FINAL MODEL\n",
      "                          MNLogit Regression Results                          \n",
      "==============================================================================\n",
      "Dep. Variable:           chosen_mode3   No. Observations:                15982\n",
      "Model:                        MNLogit   Df Residuals:                    15976\n",
      "Method:                           MLE   Df Model:                            5\n",
      "Date:                Wed, 20 Jul 2016   Pseudo R-squ.:                 0.05274\n",
      "Time:                        17:55:38   Log-Likelihood:                -1701.5\n",
      "converged:                       True   LL-Null:                       -1796.3\n",
      "                                        LLR p-value:                 5.073e-39\n",
      "==========================================================================================\n",
      "     chosen_mode3=uber       coef    std err          z      P>|z|      [95.0% Conf. Int.]\n",
      "------------------------------------------------------------------------------------------\n",
      "n_transit                 -0.5681      0.058     -9.803      0.000        -0.682    -0.455\n",
      "ln_walkwait                0.1519      0.022      6.820      0.000         0.108     0.196\n",
      "start_or_end_near_home     0.2263      0.110      2.061      0.039         0.011     0.442\n",
      "start_or_end_near_work     0.6092      0.105      5.778      0.000         0.403     0.816\n",
      "late_night                 0.4785      0.192      2.489      0.013         0.102     0.855\n",
      "const                     -3.2379      0.197    -16.423      0.000        -3.624    -2.851\n",
      "==========================================================================================\n",
      "SERIES 6 FINAL MODEL\n",
      "                          MNLogit Regression Results                          \n",
      "==============================================================================\n",
      "Dep. Variable:           chosen_mode3   No. Observations:                16031\n",
      "Model:                        MNLogit   Df Residuals:                    16025\n",
      "Method:                           MLE   Df Model:                            5\n",
      "Date:                Wed, 20 Jul 2016   Pseudo R-squ.:                 0.05155\n",
      "Time:                        17:55:38   Log-Likelihood:                -1708.3\n",
      "converged:                       True   LL-Null:                       -1801.2\n",
      "                                        LLR p-value:                 3.235e-38\n",
      "==========================================================================================\n",
      "     chosen_mode3=uber       coef    std err          z      P>|z|      [95.0% Conf. Int.]\n",
      "------------------------------------------------------------------------------------------\n",
      "n_transit                 -0.5596      0.058     -9.654      0.000        -0.673    -0.446\n",
      "walkwait                   0.0007      0.000      6.733      0.000         0.001     0.001\n",
      "start_or_end_near_home     0.2262      0.110      2.065      0.039         0.011     0.441\n",
      "start_or_end_near_work     0.5979      0.105      5.678      0.000         0.392     0.804\n",
      "late_night                 0.5189      0.192      2.703      0.007         0.143     0.895\n",
      "const                     -2.9255      0.178    -16.419      0.000        -3.275    -2.576\n",
      "==========================================================================================\n"
     ]
    }
   ],
   "source": [
    "# FINAL MODELS:\n",
    "# 1.\n",
    "exog_cols = ['n_transit','wait','transit_cost']\n",
    "defn = subset2.dropna(subset=exog_cols)\n",
    "y = defn['chosen_mode3']; x = defn.loc[:,exog_cols]; x['const'] = 1\n",
    "s1_mdl = st.MNLogit(y, x) \n",
    "s1_fit = s1_mdl.fit(maxiter=100)\n",
    "# 2.\n",
    "exog_cols = ['n_transit','wait',\n",
    "             'start_is_near_work','end_is_near_work']\n",
    "defn = subset2.dropna(subset=exog_cols)\n",
    "y = defn['chosen_mode3']; x = defn.loc[:,exog_cols]; x['const'] = 1\n",
    "s2_mdl = st.MNLogit(y, x) \n",
    "s2_fit = s2_mdl.fit(maxiter=100)\n",
    "# 3.\n",
    "exog_cols = ['n_transit','walkwait',\n",
    "             'start_or_end_near_home','start_or_end_near_work']\n",
    "defn = subset2.dropna(subset=exog_cols)\n",
    "y = defn['chosen_mode3']; x = defn.loc[:,exog_cols]; x['const'] = 1\n",
    "s3_mdl = st.MNLogit(y, x) \n",
    "s3_fit = s3_mdl.fit(maxiter=100)\n",
    "# 4.\n",
    "exog_cols = ['n_transit','ln_walkwait',\n",
    "             'start_or_end_near_home','start_or_end_near_work']\n",
    "defn = subset2.dropna(subset=exog_cols)\n",
    "y = defn['chosen_mode3']; x = defn.loc[:,exog_cols]; x['const'] = 1\n",
    "s4_mdl = st.MNLogit(y, x) \n",
    "s4_fit = s4_mdl.fit(maxiter=100)\n",
    "# 5.\n",
    "exog_cols = ['n_transit','ln_walkwait',\n",
    "             'start_or_end_near_home','start_or_end_near_work','late_night']\n",
    "defn = subset2.dropna(subset=exog_cols)\n",
    "y = defn['chosen_mode3']; x = defn.loc[:,exog_cols]; x['const'] = 1\n",
    "s5_mdl = st.MNLogit(y, x) \n",
    "s5_fit = s5_mdl.fit(maxiter=100)\n",
    "# 6.\n",
    "exog_cols = ['n_transit','walkwait',\n",
    "             'start_or_end_near_home','start_or_end_near_work','late_night']\n",
    "defn = subset2.dropna(subset=exog_cols)\n",
    "y = defn['chosen_mode3']; x = defn.loc[:,exog_cols]; x['const'] = 1\n",
    "s6_mdl = st.MNLogit(y, x) \n",
    "s6_fit = s6_mdl.fit(maxiter=100)\n",
    "\n",
    "print \"SERIES 1 FINAL MODEL\"\n",
    "print s1_fit.summary()\n",
    "print \"SERIES 2 FINAL MODEL\"\n",
    "print s2_fit.summary()\n",
    "print \"SERIES 3 FINAL MODEL\"\n",
    "print s3_fit.summary()\n",
    "print \"SERIES 4 FINAL MODEL\"\n",
    "print s4_fit.summary()\n",
    "print \"SERIES 5 FINAL MODEL\"\n",
    "print s5_fit.summary()\n",
    "print \"SERIES 6 FINAL MODEL\"\n",
    "print s6_fit.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 97,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Optimization terminated successfully.\n",
      "         Current function value: 0.106366\n",
      "         Iterations 8\n",
      "                          MNLogit Regression Results                          \n",
      "==============================================================================\n",
      "Dep. Variable:           chosen_mode3   No. Observations:                18207\n",
      "Model:                        MNLogit   Df Residuals:                    18201\n",
      "Method:                           MLE   Df Model:                            5\n",
      "Date:                Thu, 21 Jul 2016   Pseudo R-squ.:                 0.04687\n",
      "Time:                        08:19:55   Log-Likelihood:                -1936.6\n",
      "converged:                       True   LL-Null:                       -2031.8\n",
      "                                        LLR p-value:                 3.080e-39\n",
      "==========================================================================================\n",
      "     chosen_mode3=uber       coef    std err          z      P>|z|      [95.0% Conf. Int.]\n",
      "------------------------------------------------------------------------------------------\n",
      "n_transit                 -0.5647      0.054    -10.364      0.000        -0.671    -0.458\n",
      "late_night                 0.4928      0.171      2.889      0.004         0.158     0.827\n",
      "transit_time               0.0052      0.002      2.353      0.019         0.001     0.010\n",
      "wait                       0.0007      0.000      6.504      0.000         0.000     0.001\n",
      "start_or_end_near_home     0.2055      0.104      1.983      0.047         0.002     0.409\n",
      "const                     -2.7931      0.175    -16.000      0.000        -3.135    -2.451\n",
      "==========================================================================================\n"
     ]
    }
   ],
   "source": [
    "# --------------------------\n",
    "# - SERIES 7               - \n",
    "# --------------------------\n",
    "# Start with n_transit and late_night\n",
    "# added variables one at a time, and only kept them if all variables remained significant\n",
    "exog_cols = ['n_transit','late_night','transit_time','wait','start_or_end_near_home']\n",
    "defn = subset2.dropna(subset=exog_cols)\n",
    "y = defn['chosen_mode3']\n",
    "x = defn.loc[:,exog_cols]\n",
    "x['const'] = 1\n",
    "\n",
    "s2_mdl = st.MNLogit(y, x)\n",
    "s2_fit = s2_mdl.fit(maxiter=100)\n",
    "print s2_fit.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 86,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Optimization terminated successfully.\n",
      "         Current function value: 0.110295\n",
      "         Iterations 8\n",
      "                          MNLogit Regression Results                          \n",
      "==============================================================================\n",
      "Dep. Variable:           chosen_mode3   No. Observations:                16254\n",
      "Model:                        MNLogit   Df Residuals:                    16249\n",
      "Method:                           MLE   Df Model:                            4\n",
      "Date:                Wed, 20 Jul 2016   Pseudo R-squ.:                 0.01372\n",
      "Time:                        18:04:51   Log-Likelihood:                -1792.7\n",
      "converged:                       True   LL-Null:                       -1817.7\n",
      "                                        LLR p-value:                 3.798e-10\n",
      "==========================================================================================\n",
      "     chosen_mode3=uber       coef    std err          z      P>|z|      [95.0% Conf. Int.]\n",
      "------------------------------------------------------------------------------------------\n",
      "late_night                 0.6245      0.190      3.289      0.001         0.252     0.997\n",
      "uber_basic_cost            0.0342      0.011      3.163      0.002         0.013     0.055\n",
      "uber_basic_time           -0.0174      0.010     -1.680      0.093        -0.038     0.003\n",
      "start_or_end_near_work     0.5611      0.106      5.289      0.000         0.353     0.769\n",
      "const                     -4.1587      0.105    -39.507      0.000        -4.365    -3.952\n",
      "==========================================================================================\n"
     ]
    }
   ],
   "source": [
    "# --------------------------\n",
    "# - SERIES 7               - \n",
    "# --------------------------\n",
    "# based mainly on Uber attributes\n",
    "exog_cols = ['late_night','start_or_end_near_work']\n",
    "defn = subset2.dropna(subset=exog_cols)\n",
    "y = defn['chosen_mode3']\n",
    "x = defn.loc[:,exog_cols]\n",
    "x['const'] = 1\n",
    "\n",
    "s2_mdl = st.MNLogit(y, x)\n",
    "s2_fit = s2_mdl.fit(maxiter=500)\n",
    "print s2_fit.summary()"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 2",
   "language": "python",
   "name": "python2"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 2
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython2",
   "version": "2.7.7"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 0
}