EDA Feature Engineering & LightGBM Modelling.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Author: Chongzheng Zhao\n",
    "# Kaggle Competition - New York City Taxi Fare Prediction\n",
    "# FINAL BEAT 1454 TEAMS!\n",
    "# We team NYCTAXI located at 30 out of 1484 teams, Top 2%!\n",
    "# Final Score 2.86770\n",
    "# Competition Official Website: https://www.kaggle.com/c/new-york-city-taxi-fare-prediction\n",
    "# To see the leaderboard(competition result): https://www.kaggle.com/c/new-york-city-taxi-fare-prediction/leaderboard\n",
    "# Welcome to my Github profile: https://github.com/ChongzhengZhao/\n",
    "# Welcome to my Kaggle Profile: https://www.kaggle.com/chongzhengzhao\n",
    "# Welcome to my Linkedin Profile: https://www.linkedin.com/in/chongzhengzhao/\n",
    "# Last updated: 30/11/2018"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import pandas as pd\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "% matplotlib inline\n",
    "plt.style.use('seaborn-whitegrid')\n",
    "import warnings\n",
    "warnings.filterwarnings('ignore')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "train = pd.read_csv(\"train.csv\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(55423856, 8)"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "key                   object\n",
       "fare_amount          float64\n",
       "pickup_datetime       object\n",
       "pickup_longitude     float64\n",
       "pickup_latitude      float64\n",
       "dropoff_longitude    float64\n",
       "dropoff_latitude     float64\n",
       "passenger_count        int64\n",
       "dtype: object"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train.dtypes"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "key                    0\n",
       "fare_amount            0\n",
       "pickup_datetime        0\n",
       "pickup_longitude       0\n",
       "pickup_latitude        0\n",
       "dropoff_longitude    376\n",
       "dropoff_latitude     376\n",
       "passenger_count        0\n",
       "dtype: int64"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train.isnull().sum()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>fare_amount</th>\n",
       "      <th>pickup_longitude</th>\n",
       "      <th>pickup_latitude</th>\n",
       "      <th>dropoff_longitude</th>\n",
       "      <th>dropoff_latitude</th>\n",
       "      <th>passenger_count</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>count</th>\n",
       "      <td>5.542386e+07</td>\n",
       "      <td>5.542386e+07</td>\n",
       "      <td>5.542386e+07</td>\n",
       "      <td>5.542348e+07</td>\n",
       "      <td>5.542348e+07</td>\n",
       "      <td>5.542386e+07</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mean</th>\n",
       "      <td>1.134505e+01</td>\n",
       "      <td>-7.250968e+01</td>\n",
       "      <td>3.991979e+01</td>\n",
       "      <td>-7.251121e+01</td>\n",
       "      <td>3.992068e+01</td>\n",
       "      <td>1.685380e+00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>std</th>\n",
       "      <td>2.071083e+01</td>\n",
       "      <td>1.284888e+01</td>\n",
       "      <td>9.642353e+00</td>\n",
       "      <td>1.278220e+01</td>\n",
       "      <td>9.633346e+00</td>\n",
       "      <td>1.327664e+00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>min</th>\n",
       "      <td>-3.000000e+02</td>\n",
       "      <td>-3.442060e+03</td>\n",
       "      <td>-3.492264e+03</td>\n",
       "      <td>-3.442025e+03</td>\n",
       "      <td>-3.547887e+03</td>\n",
       "      <td>0.000000e+00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25%</th>\n",
       "      <td>6.000000e+00</td>\n",
       "      <td>-7.399207e+01</td>\n",
       "      <td>4.073493e+01</td>\n",
       "      <td>-7.399140e+01</td>\n",
       "      <td>4.073403e+01</td>\n",
       "      <td>1.000000e+00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>50%</th>\n",
       "      <td>8.500000e+00</td>\n",
       "      <td>-7.398180e+01</td>\n",
       "      <td>4.075265e+01</td>\n",
       "      <td>-7.398015e+01</td>\n",
       "      <td>4.075316e+01</td>\n",
       "      <td>1.000000e+00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>75%</th>\n",
       "      <td>1.250000e+01</td>\n",
       "      <td>-7.396708e+01</td>\n",
       "      <td>4.076713e+01</td>\n",
       "      <td>-7.396367e+01</td>\n",
       "      <td>4.076810e+01</td>\n",
       "      <td>2.000000e+00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>max</th>\n",
       "      <td>9.396336e+04</td>\n",
       "      <td>3.457626e+03</td>\n",
       "      <td>3.408790e+03</td>\n",
       "      <td>3.457622e+03</td>\n",
       "      <td>3.537133e+03</td>\n",
       "      <td>2.080000e+02</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "        fare_amount  pickup_longitude  pickup_latitude  dropoff_longitude  \\\n",
       "count  5.542386e+07      5.542386e+07     5.542386e+07       5.542348e+07   \n",
       "mean   1.134505e+01     -7.250968e+01     3.991979e+01      -7.251121e+01   \n",
       "std    2.071083e+01      1.284888e+01     9.642353e+00       1.278220e+01   \n",
       "min   -3.000000e+02     -3.442060e+03    -3.492264e+03      -3.442025e+03   \n",
       "25%    6.000000e+00     -7.399207e+01     4.073493e+01      -7.399140e+01   \n",
       "50%    8.500000e+00     -7.398180e+01     4.075265e+01      -7.398015e+01   \n",
       "75%    1.250000e+01     -7.396708e+01     4.076713e+01      -7.396367e+01   \n",
       "max    9.396336e+04      3.457626e+03     3.408790e+03       3.457622e+03   \n",
       "\n",
       "       dropoff_latitude  passenger_count  \n",
       "count      5.542348e+07     5.542386e+07  \n",
       "mean       3.992068e+01     1.685380e+00  \n",
       "std        9.633346e+00     1.327664e+00  \n",
       "min       -3.547887e+03     0.000000e+00  \n",
       "25%        4.073403e+01     1.000000e+00  \n",
       "50%        4.075316e+01     1.000000e+00  \n",
       "75%        4.076810e+01     2.000000e+00  \n",
       "max        3.537133e+03     2.080000e+02  "
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train.describe()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "#Fare amount is negative and it doesn't seem to be realistic\n",
    "#few longitude and lattitude entries are off\n",
    "#maximum passanger count is 208 which looks odd"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "train.dropna(how='any', axis=0, inplace = True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(55423480, 8)"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train.shape #drop 376 "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "train = train[train.fare_amount >=0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(55421026, 8)"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train.shape # drop 2454"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(101, 8)"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train[train.passenger_count >7].shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "train=train[train.passenger_count <=7]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(55420925, 8)"
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train.shape #drop 101"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(195039, 8)"
      ]
     },
     "execution_count": 19,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train[train.passenger_count ==0].shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "#Passanger count is incorrectly populated\n",
    "#Taxi was not carrying any passanger, may be taxi was used for goods"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "train=train[train.passenger_count >0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(55225886, 8)"
      ]
     },
     "execution_count": 22,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train.shape # drop 195k"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "test = pd.read_csv(\"test.csv\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(9914, 7)"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "test.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>pickup_longitude</th>\n",
       "      <th>pickup_latitude</th>\n",
       "      <th>dropoff_longitude</th>\n",
       "      <th>dropoff_latitude</th>\n",
       "      <th>passenger_count</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>count</th>\n",
       "      <td>9914.000000</td>\n",
       "      <td>9914.000000</td>\n",
       "      <td>9914.000000</td>\n",
       "      <td>9914.000000</td>\n",
       "      <td>9914.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mean</th>\n",
       "      <td>-73.974722</td>\n",
       "      <td>40.751041</td>\n",
       "      <td>-73.973657</td>\n",
       "      <td>40.751743</td>\n",
       "      <td>1.671273</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>std</th>\n",
       "      <td>0.042774</td>\n",
       "      <td>0.033541</td>\n",
       "      <td>0.039072</td>\n",
       "      <td>0.035435</td>\n",
       "      <td>1.278747</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>min</th>\n",
       "      <td>-74.252193</td>\n",
       "      <td>40.573143</td>\n",
       "      <td>-74.263242</td>\n",
       "      <td>40.568973</td>\n",
       "      <td>1.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25%</th>\n",
       "      <td>-73.992501</td>\n",
       "      <td>40.736125</td>\n",
       "      <td>-73.991247</td>\n",
       "      <td>40.735254</td>\n",
       "      <td>1.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>50%</th>\n",
       "      <td>-73.982326</td>\n",
       "      <td>40.753051</td>\n",
       "      <td>-73.980015</td>\n",
       "      <td>40.754065</td>\n",
       "      <td>1.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>75%</th>\n",
       "      <td>-73.968013</td>\n",
       "      <td>40.767113</td>\n",
       "      <td>-73.964059</td>\n",
       "      <td>40.768757</td>\n",
       "      <td>2.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>max</th>\n",
       "      <td>-72.986532</td>\n",
       "      <td>41.709555</td>\n",
       "      <td>-72.990963</td>\n",
       "      <td>41.696683</td>\n",
       "      <td>6.000000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       pickup_longitude  pickup_latitude  dropoff_longitude  dropoff_latitude  \\\n",
       "count       9914.000000      9914.000000        9914.000000       9914.000000   \n",
       "mean         -73.974722        40.751041         -73.973657         40.751743   \n",
       "std            0.042774         0.033541           0.039072          0.035435   \n",
       "min          -74.252193        40.573143         -74.263242         40.568973   \n",
       "25%          -73.992501        40.736125         -73.991247         40.735254   \n",
       "50%          -73.982326        40.753051         -73.980015         40.754065   \n",
       "75%          -73.968013        40.767113         -73.964059         40.768757   \n",
       "max          -72.986532        41.709555         -72.990963         41.696683   \n",
       "\n",
       "       passenger_count  \n",
       "count      9914.000000  \n",
       "mean          1.671273  \n",
       "std           1.278747  \n",
       "min           1.000000  \n",
       "25%           1.000000  \n",
       "50%           1.000000  \n",
       "75%           2.000000  \n",
       "max           6.000000  "
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "test.describe()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "#store the minimum and maximum of the longitude and latitude from test data set and filter the train data set for those data points"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(-74.263242000000005, -72.986531999999997)"
      ]
     },
     "execution_count": 24,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "min(test.pickup_longitude.min(),test.dropoff_longitude.min()), \\\n",
    "max(test.pickup_longitude.max(),test.dropoff_longitude.max())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(40.568973, 41.709555000000002)"
      ]
     },
     "execution_count": 25,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "min(test.pickup_latitude.min(),test.dropoff_latitude.min()), \\\n",
    "max(test.pickup_latitude.max(),test.dropoff_latitude.max())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def select_within_test_boundary(df, BB):\n",
    "    return (df.pickup_longitude >= BB[0]) & (df.pickup_longitude <= BB[1]) & \\\n",
    "           (df.pickup_latitude >= BB[2]) & (df.pickup_latitude <= BB[3]) & \\\n",
    "           (df.dropoff_longitude >= BB[0]) & (df.dropoff_longitude <= BB[1]) & \\\n",
    "           (df.dropoff_latitude >= BB[2]) & (df.dropoff_latitude <= BB[3])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(54055894, 8)"
      ]
     },
     "execution_count": 27,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "BB = (-74.5, -72.8, 40.5, 41.7)\n",
    "train = train[select_within_test_boundary(train, BB)]\n",
    "train.shape # drop 1,169,992"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>key</th>\n",
       "      <th>fare_amount</th>\n",
       "      <th>pickup_datetime</th>\n",
       "      <th>pickup_longitude</th>\n",
       "      <th>pickup_latitude</th>\n",
       "      <th>dropoff_longitude</th>\n",
       "      <th>dropoff_latitude</th>\n",
       "      <th>passenger_count</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>2009-06-15 17:26:21.0000001</td>\n",
       "      <td>4.5</td>\n",
       "      <td>2009-06-15 17:26:21 UTC</td>\n",
       "      <td>-73.844311</td>\n",
       "      <td>40.721319</td>\n",
       "      <td>-73.841610</td>\n",
       "      <td>40.712278</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2010-01-05 16:52:16.0000002</td>\n",
       "      <td>16.9</td>\n",
       "      <td>2010-01-05 16:52:16 UTC</td>\n",
       "      <td>-74.016048</td>\n",
       "      <td>40.711303</td>\n",
       "      <td>-73.979268</td>\n",
       "      <td>40.782004</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2011-08-18 00:35:00.00000049</td>\n",
       "      <td>5.7</td>\n",
       "      <td>2011-08-18 00:35:00 UTC</td>\n",
       "      <td>-73.982738</td>\n",
       "      <td>40.761270</td>\n",
       "      <td>-73.991242</td>\n",
       "      <td>40.750562</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>2012-04-21 04:30:42.0000001</td>\n",
       "      <td>7.7</td>\n",
       "      <td>2012-04-21 04:30:42 UTC</td>\n",
       "      <td>-73.987130</td>\n",
       "      <td>40.733143</td>\n",
       "      <td>-73.991567</td>\n",
       "      <td>40.758092</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>2010-03-09 07:51:00.000000135</td>\n",
       "      <td>5.3</td>\n",
       "      <td>2010-03-09 07:51:00 UTC</td>\n",
       "      <td>-73.968095</td>\n",
       "      <td>40.768008</td>\n",
       "      <td>-73.956655</td>\n",
       "      <td>40.783762</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                             key  fare_amount          pickup_datetime  \\\n",
       "0    2009-06-15 17:26:21.0000001          4.5  2009-06-15 17:26:21 UTC   \n",
       "1    2010-01-05 16:52:16.0000002         16.9  2010-01-05 16:52:16 UTC   \n",
       "2   2011-08-18 00:35:00.00000049          5.7  2011-08-18 00:35:00 UTC   \n",
       "3    2012-04-21 04:30:42.0000001          7.7  2012-04-21 04:30:42 UTC   \n",
       "4  2010-03-09 07:51:00.000000135          5.3  2010-03-09 07:51:00 UTC   \n",
       "\n",
       "   pickup_longitude  pickup_latitude  dropoff_longitude  dropoff_latitude  \\\n",
       "0        -73.844311        40.721319         -73.841610         40.712278   \n",
       "1        -74.016048        40.711303         -73.979268         40.782004   \n",
       "2        -73.982738        40.761270         -73.991242         40.750562   \n",
       "3        -73.987130        40.733143         -73.991567         40.758092   \n",
       "4        -73.968095        40.768008         -73.956655         40.783762   \n",
       "\n",
       "   passenger_count  \n",
       "0                1  \n",
       "1                1  \n",
       "2                2  \n",
       "3                1  \n",
       "4                1  "
      ]
     },
     "execution_count": 28,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train.head() "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>key</th>\n",
       "      <th>fare_amount</th>\n",
       "      <th>pickup_datetime</th>\n",
       "      <th>pickup_longitude</th>\n",
       "      <th>pickup_latitude</th>\n",
       "      <th>dropoff_longitude</th>\n",
       "      <th>dropoff_latitude</th>\n",
       "      <th>passenger_count</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>55423851</th>\n",
       "      <td>2014-03-15 03:28:00.00000070</td>\n",
       "      <td>14.0</td>\n",
       "      <td>2014-03-15 03:28:00 UTC</td>\n",
       "      <td>-74.005272</td>\n",
       "      <td>40.740027</td>\n",
       "      <td>-73.963280</td>\n",
       "      <td>40.762555</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>55423852</th>\n",
       "      <td>2009-03-24 20:46:20.0000002</td>\n",
       "      <td>4.2</td>\n",
       "      <td>2009-03-24 20:46:20 UTC</td>\n",
       "      <td>-73.957784</td>\n",
       "      <td>40.765530</td>\n",
       "      <td>-73.951640</td>\n",
       "      <td>40.773959</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>55423853</th>\n",
       "      <td>2011-04-02 22:04:24.0000004</td>\n",
       "      <td>14.1</td>\n",
       "      <td>2011-04-02 22:04:24 UTC</td>\n",
       "      <td>-73.970505</td>\n",
       "      <td>40.752325</td>\n",
       "      <td>-73.960537</td>\n",
       "      <td>40.797342</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>55423854</th>\n",
       "      <td>2011-10-26 05:57:51.0000002</td>\n",
       "      <td>28.9</td>\n",
       "      <td>2011-10-26 05:57:51 UTC</td>\n",
       "      <td>-73.980901</td>\n",
       "      <td>40.764629</td>\n",
       "      <td>-73.870605</td>\n",
       "      <td>40.773963</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>55423855</th>\n",
       "      <td>2014-12-12 11:33:00.00000015</td>\n",
       "      <td>7.5</td>\n",
       "      <td>2014-12-12 11:33:00 UTC</td>\n",
       "      <td>-73.969722</td>\n",
       "      <td>40.797668</td>\n",
       "      <td>-73.970885</td>\n",
       "      <td>40.783313</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                   key  fare_amount          pickup_datetime  \\\n",
       "55423851  2014-03-15 03:28:00.00000070         14.0  2014-03-15 03:28:00 UTC   \n",
       "55423852   2009-03-24 20:46:20.0000002          4.2  2009-03-24 20:46:20 UTC   \n",
       "55423853   2011-04-02 22:04:24.0000004         14.1  2011-04-02 22:04:24 UTC   \n",
       "55423854   2011-10-26 05:57:51.0000002         28.9  2011-10-26 05:57:51 UTC   \n",
       "55423855  2014-12-12 11:33:00.00000015          7.5  2014-12-12 11:33:00 UTC   \n",
       "\n",
       "          pickup_longitude  pickup_latitude  dropoff_longitude  \\\n",
       "55423851        -74.005272        40.740027         -73.963280   \n",
       "55423852        -73.957784        40.765530         -73.951640   \n",
       "55423853        -73.970505        40.752325         -73.960537   \n",
       "55423854        -73.980901        40.764629         -73.870605   \n",
       "55423855        -73.969722        40.797668         -73.970885   \n",
       "\n",
       "          dropoff_latitude  passenger_count  \n",
       "55423851         40.762555                1  \n",
       "55423852         40.773959                1  \n",
       "55423853         40.797342                1  \n",
       "55423854         40.773963                1  \n",
       "55423855         40.783313                1  "
      ]
     },
     "execution_count": 29,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train.tail()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>key</th>\n",
       "      <th>pickup_datetime</th>\n",
       "      <th>pickup_longitude</th>\n",
       "      <th>pickup_latitude</th>\n",
       "      <th>dropoff_longitude</th>\n",
       "      <th>dropoff_latitude</th>\n",
       "      <th>passenger_count</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>2015-01-27 13:08:24.0000002</td>\n",
       "      <td>2015-01-27 13:08:24 UTC</td>\n",
       "      <td>-73.973320</td>\n",
       "      <td>40.763805</td>\n",
       "      <td>-73.981430</td>\n",
       "      <td>40.743835</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2015-01-27 13:08:24.0000003</td>\n",
       "      <td>2015-01-27 13:08:24 UTC</td>\n",
       "      <td>-73.986862</td>\n",
       "      <td>40.719383</td>\n",
       "      <td>-73.998886</td>\n",
       "      <td>40.739201</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2011-10-08 11:53:44.0000002</td>\n",
       "      <td>2011-10-08 11:53:44 UTC</td>\n",
       "      <td>-73.982524</td>\n",
       "      <td>40.751260</td>\n",
       "      <td>-73.979654</td>\n",
       "      <td>40.746139</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>2012-12-01 21:12:12.0000002</td>\n",
       "      <td>2012-12-01 21:12:12 UTC</td>\n",
       "      <td>-73.981160</td>\n",
       "      <td>40.767807</td>\n",
       "      <td>-73.990448</td>\n",
       "      <td>40.751635</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>2012-12-01 21:12:12.0000003</td>\n",
       "      <td>2012-12-01 21:12:12 UTC</td>\n",
       "      <td>-73.966046</td>\n",
       "      <td>40.789775</td>\n",
       "      <td>-73.988565</td>\n",
       "      <td>40.744427</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                           key          pickup_datetime  pickup_longitude  \\\n",
       "0  2015-01-27 13:08:24.0000002  2015-01-27 13:08:24 UTC        -73.973320   \n",
       "1  2015-01-27 13:08:24.0000003  2015-01-27 13:08:24 UTC        -73.986862   \n",
       "2  2011-10-08 11:53:44.0000002  2011-10-08 11:53:44 UTC        -73.982524   \n",
       "3  2012-12-01 21:12:12.0000002  2012-12-01 21:12:12 UTC        -73.981160   \n",
       "4  2012-12-01 21:12:12.0000003  2012-12-01 21:12:12 UTC        -73.966046   \n",
       "\n",
       "   pickup_latitude  dropoff_longitude  dropoff_latitude  passenger_count  \n",
       "0        40.763805         -73.981430         40.743835                1  \n",
       "1        40.719383         -73.998886         40.739201                1  \n",
       "2        40.751260         -73.979654         40.746139                1  \n",
       "3        40.767807         -73.990448         40.751635                1  \n",
       "4        40.789775         -73.988565         40.744427                1  "
      ]
     },
     "execution_count": 30,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "test.head() "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>key</th>\n",
       "      <th>pickup_datetime</th>\n",
       "      <th>pickup_longitude</th>\n",
       "      <th>pickup_latitude</th>\n",
       "      <th>dropoff_longitude</th>\n",
       "      <th>dropoff_latitude</th>\n",
       "      <th>passenger_count</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>9909</th>\n",
       "      <td>2015-05-10 12:37:51.0000002</td>\n",
       "      <td>2015-05-10 12:37:51 UTC</td>\n",
       "      <td>-73.968124</td>\n",
       "      <td>40.796997</td>\n",
       "      <td>-73.955643</td>\n",
       "      <td>40.780388</td>\n",
       "      <td>6</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9910</th>\n",
       "      <td>2015-01-12 17:05:51.0000001</td>\n",
       "      <td>2015-01-12 17:05:51 UTC</td>\n",
       "      <td>-73.945511</td>\n",
       "      <td>40.803600</td>\n",
       "      <td>-73.960213</td>\n",
       "      <td>40.776371</td>\n",
       "      <td>6</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9911</th>\n",
       "      <td>2015-04-19 20:44:15.0000001</td>\n",
       "      <td>2015-04-19 20:44:15 UTC</td>\n",
       "      <td>-73.991600</td>\n",
       "      <td>40.726608</td>\n",
       "      <td>-73.789742</td>\n",
       "      <td>40.647011</td>\n",
       "      <td>6</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9912</th>\n",
       "      <td>2015-01-31 01:05:19.0000005</td>\n",
       "      <td>2015-01-31 01:05:19 UTC</td>\n",
       "      <td>-73.985573</td>\n",
       "      <td>40.735432</td>\n",
       "      <td>-73.939178</td>\n",
       "      <td>40.801731</td>\n",
       "      <td>6</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9913</th>\n",
       "      <td>2015-01-18 14:06:23.0000006</td>\n",
       "      <td>2015-01-18 14:06:23 UTC</td>\n",
       "      <td>-73.988022</td>\n",
       "      <td>40.754070</td>\n",
       "      <td>-74.000282</td>\n",
       "      <td>40.759220</td>\n",
       "      <td>6</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                              key          pickup_datetime  pickup_longitude  \\\n",
       "9909  2015-05-10 12:37:51.0000002  2015-05-10 12:37:51 UTC        -73.968124   \n",
       "9910  2015-01-12 17:05:51.0000001  2015-01-12 17:05:51 UTC        -73.945511   \n",
       "9911  2015-04-19 20:44:15.0000001  2015-04-19 20:44:15 UTC        -73.991600   \n",
       "9912  2015-01-31 01:05:19.0000005  2015-01-31 01:05:19 UTC        -73.985573   \n",
       "9913  2015-01-18 14:06:23.0000006  2015-01-18 14:06:23 UTC        -73.988022   \n",
       "\n",
       "      pickup_latitude  dropoff_longitude  dropoff_latitude  passenger_count  \n",
       "9909        40.796997         -73.955643         40.780388                6  \n",
       "9910        40.803600         -73.960213         40.776371                6  \n",
       "9911        40.726608         -73.789742         40.647011                6  \n",
       "9912        40.735432         -73.939178         40.801731                6  \n",
       "9913        40.754070         -74.000282         40.759220                6  "
      ]
     },
     "execution_count": 31,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "test.tail() "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# eda sample 2,000,000 obs"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "train_sample = train.sample(n=2000000)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(2000000, 8)"
      ]
     },
     "execution_count": 34,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_sample.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.text.Text at 0x458f1ef10>"
      ]
     },
     "execution_count": 35,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAA8cAAAHfCAYAAAB9FaqFAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x456a056d0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Lets see the distribution of fare amount \n",
    "train_sample.fare_amount.hist(bins=100, figsize = (16,8))\n",
    "plt.xlabel(\"Fare Amount\")\n",
    "plt.ylabel(\"Frequency\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.text.Text at 0x460a476d0>"
      ]
     },
     "execution_count": 36,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAA8EAAAHfCAYAAABwC9rCAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x457653c10>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Lets see the distribution of fare amount less than 100\n",
    "train_sample[train_sample.fare_amount <=100 ].fare_amount.hist(bins=100, figsize = (16,8))\n",
    "plt.xlabel(\"Fare Amount\")\n",
    "plt.ylabel(\"Frequency\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "#There are few points between 40 and 60 dollars which has slightly high frequency and that could be airport trips"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.text.Text at 0x39fdde650>"
      ]
     },
     "execution_count": 38,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAA6sAAAHfCAYAAAC2zk57AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\niiNWAQAAKI5YBQAAoDhiFQAAgOL8P8vE4kGGUW0jAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x39f667050>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "train_sample[train_sample.fare_amount >100 ].fare_amount.hist(bins=100, figsize = (16,8))\n",
    "plt.xlabel(\"Fare Amount\")\n",
    "plt.ylabel(\"Frequency\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "#some of them might be outliers or few of them might be long distance trip from/to airport"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.text.Text at 0x39e844e10>"
      ]
     },
     "execution_count": 39,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAA8cAAAHfCAYAAAB9FaqFAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAAIABJREFUeJzs3X+cl3Wd7//nB2YYbGYIydzNcjphknq8gfzQbHegLE/o\nnkwj5Md0qNZyj9wUhZt6UFSIyojjMtsKh/xxoz0tCoS/Uo9lraYg4kHPnAValDZdfyHqUbGNmYDB\nmc/3j70130iHZtP5Adf9/tfMNddc1+uaz1vr4XXNZ0rlcrkcAAAAKLB+vT0AAAAA9DZxDAAAQOGJ\nYwAAAApPHAMAAFB44hgAAIDCE8cAAAAUXkVvD9DXNDU19fYIAAAAdJPRo0e/5XZx/BY6+2H1BU1N\nTX16PnqPtcH+WB90xtqgM9YG+2N90Jm+vjb2dzPUY9UAAAAUnjgGAACg8MQxAAAAhSeOAQAAKDxx\nDAAAQOGJYwAAAApPHAMAAFB44hgAAIDCE8cAAAAUnjgGAACg8MQxAAAAhSeOAQAAKLxujeNNmzZl\n2rRp+2y7++67M3ny5I7PV69enQkTJmTSpEl54IEHkiS7d+/OjBkz0tDQkHPPPTc7duxIkmzcuDFn\nn312pkyZkiVLlnQcY8mSJZk4cWKmTJmSzZs3J0l27NiRc845Jw0NDZk5c2Z27drVnZcKAADAAazb\n4vjGG2/MlVdemT179nRse/zxx3PrrbemXC4nSV555ZUsX748q1atyrJly9LY2JjW1tasXLkyw4YN\ny4oVK3LWWWdl6dKlSZJ58+Zl0aJFWblyZTZt2pTHH388W7ZsyaOPPppbbrkljY2NmT9/fpJk6dKl\n+cxnPpMVK1bkuOOOyw9+8IPuulQAAAAOcN0Wx3V1dVm8eHHH56+//noaGxszZ86cjm2bN2/OyJEj\nM2DAgNTW1qauri5bt25NU1NTxo4dmyQZN25cHnnkkTQ3N6e1tTV1dXUplUqpr6/P+vXr09TUlPr6\n+pRKpRxxxBFpa2vLjh073nSM9evXd9elAgAAcICr6K4Djx8/Ptu2bUuStLW15Yorrsjll1+eqqqq\njn2am5tTW1vb8Xl1dXWam5v32V5dXZ2dO3emubk5NTU1++z7/PPPp6qqKoMHD95n+2/3//1jdFVT\nU9Mfd9E9pK/PR++xNtgf64POWBt0xtpgf6wPOnOgro1ui+PftWXLljz77LP52te+lj179uTJJ5/M\n1VdfnZNPPjktLS0d+7W0tKS2tjY1NTUd21taWjJo0KB9tv3u9srKyv0eY+DAgR37dtXo0aPfgavu\nHk1NTX16PnqPtcH+WB90xtqgM9YG+2N90Jm+vjb2F+498m7Vw4cPzz333JPly5ensbExH/7wh3PF\nFVdk+PDhaWpqyp49e7Jz58489dRTGTZsWEaNGpU1a9YkSdauXZvRo0enpqYmlZWVee6551Iul7Nu\n3bqMGTMmo0aNyrp169Le3p7t27envb09Q4YMectjAAAAwFvpkTvHnXnve9+badOmpaGhIeVyObNm\nzUpVVVWmTp2a2bNnZ+rUqamsrMyiRYuSJPPnz88ll1yStra21NfXZ8SIEUmSMWPGZPLkyWlvb8/c\nuXOTJNOnT8/s2bOzevXqHHrooR3HAAAAgN9XKv/2raNJ0vcfAzjj4jt7e4QD3t2LzuztEbpFX1+7\n9C7rg85YG3TG2mB/rA8609fXxv7m65HHqgEAAKAvE8cAAAAUnjgGAACg8MQxAAAAhSeOAQAAKDxx\nDAAAQOGJYwAAAApPHAMAAFB44hgAAIDCE8cAAAAUnjgGAACg8MQxAAAAhSeOAQAAKDxxDAAAQOGJ\nYwAAAApPHAMAAFB44hgAAIDCE8cAAAAUnjgGAACg8MQxAAAAhSeOAQAAKDxxDAAAQOGJYwAAAApP\nHAMAAFB44hgAAIDCE8cAAAAUnjgGAACg8MQxAAAAhSeOAQAAKDxxDAAAQOGJYwAAAApPHAMAAFB4\n4hgAAIDCE8cAAAAUnjgGAACg8MQxAAAAhSeOAQAAKDxxDAAAQOGJYwAAAApPHAMAAFB44hgAAIDC\nE8cAAAAUnjgGAACg8MQxAAAAhSeOAQAAKDxxDAAAQOGJYwAAAApPHAMAAFB44hgAAIDCE8cAAAAU\nnjgGAACg8MQxAAAAhdetcbxp06ZMmzYtSfLEE0+koaEh06ZNy1e+8pW8+uqrSZLVq1dnwoQJmTRp\nUh544IEkye7duzNjxow0NDTk3HPPzY4dO5IkGzduzNlnn50pU6ZkyZIlHedZsmRJJk6cmClTpmTz\n5s1Jkh07duScc85JQ0NDZs6cmV27dnXnpQIAAHAA67Y4vvHGG3PllVdmz549SZKrr746V111VZYv\nX57/9J/+U2688ca88sorWb58eVatWpVly5alsbExra2tWblyZYYNG5YVK1bkrLPOytKlS5Mk8+bN\ny6JFi7Jy5cps2rQpjz/+eLZs2ZJHH300t9xySxobGzN//vwkydKlS/OZz3wmK1asyHHHHZcf/OAH\n3XWpAAAAHOC6LY7r6uqyePHijs8bGxtz7LHHJkna2tpSVVWVzZs3Z+TIkRkwYEBqa2tTV1eXrVu3\npqmpKWPHjk2SjBs3Lo888kiam5vT2tqaurq6lEql1NfXZ/369Wlqakp9fX1KpVKOOOKItLW1ZceO\nHW86xvr167vrUgEAADjAVXTXgcePH59t27Z1fH744YcnSf7v//2/uemmm3LzzTfnoYceSm1tbcc+\n1dXVaW5uTnNzc8f26urq7Ny5M83Nzampqdln3+effz5VVVUZPHjwPtt/u//vH6Ormpqa/riL5oBw\nML++B/O18fZZH3TG2qAz1gb7Y33QmQN1bXRbHL+VH/3oR/nud7+bG264IUOGDElNTU1aWlo6vt7S\n0pLa2tp9tre0tGTQoEFvue+gQYNSWVm532MMHDiwY9+uGj169Dtwtd1kxbY/vA/71adf37ehqanp\noL023j7rg85YG3TG2mB/rA8609fXxv7CvcferfrOO+/MTTfdlOXLl+fII49MkgwfPjxNTU3Zs2dP\ndu7cmaeeeirDhg3LqFGjsmbNmiTJ2rVrM3r06NTU1KSysjLPPfdcyuVy1q1blzFjxmTUqFFZt25d\n2tvbs3379rS3t2fIkCFveQwAAAB4Kz1y57itrS1XX3113ve+92XGjBlJkhNPPDEXXnhhpk2bloaG\nhpTL5cyaNStVVVWZOnVqZs+enalTp6aysjKLFi1KksyfPz+XXHJJ2traUl9fnxEjRiRJxowZk8mT\nJ6e9vT1z585NkkyfPj2zZ8/O6tWrc+ihh3YcAwAAAH5ft8bxBz7wgaxevTpJ8uijj77lPpMmTcqk\nSZP22XbIIYfk2muvfdO+J5xwQsfxfteMGTM6ovu3DjvssCxbtuyPHR0AAIAC6bHHqgEAAKCvEscA\nAAAUnjgGAACg8MQxAAAAhSeOAQAAKDxxDAAAQOGJYwAAAApPHAMAAFB44hgAAIDCE8cAAAAUnjgG\nAACg8MQxAAAAhSeOAQAAKDxxDAAAQOGJYwAAAApPHAMAAFB44hgAAIDCE8cAAAAUnjgGAACg8MQx\nAAAAhSeOAQAAKDxxDAAAQOGJYwAAAApPHAMAAFB44hgAAIDCE8cAAAAUnjgGAACg8MQxAAAAhSeO\nAQAAKDxxDAAAQOGJYwAAAApPHAMAAFB44hgAAIDCE8cAAAAUnjgGAACg8MQxAAAAhSeOAQAAKDxx\nDAAAQOGJYwAAAApPHAMAAFB44hgAAIDCE8cAAAAUnjgGAACg8MQxAAAAhSeOAQAAKDxxDAAAQOGJ\nYwAAAApPHAMAAFB44hgAAIDCE8cAAAAUnjgGAACg8Lo1jjdt2pRp06YlSZ599tlMnTo1DQ0NmTdv\nXtrb25Mkq1evzoQJEzJp0qQ88MADSZLdu3dnxowZaWhoyLnnnpsdO3YkSTZu3Jizzz47U6ZMyZIl\nSzrOs2TJkkycODFTpkzJ5s2bkyQ7duzIOeeck4aGhsycOTO7du3qzksFAADgANZtcXzjjTfmyiuv\nzJ49e5IkCxYsyMyZM7NixYqUy+Xcf//9eeWVV7J8+fKsWrUqy5YtS2NjY1pbW7Ny5coMGzYsK1as\nyFlnnZWlS5cmSebNm5dFixZl5cqV2bRpUx5//PFs2bIljz76aG655ZY0NjZm/vz5SZKlS5fmM5/5\nTFasWJHjjjsuP/jBD7rrUgEAADjAdVsc19XVZfHixR2fb9myJSeddFKSZNy4cVm/fn02b96ckSNH\nZsCAAamtrU1dXV22bt2apqamjB07tmPfRx55JM3NzWltbU1dXV1KpVLq6+uzfv36NDU1pb6+PqVS\nKUcccUTa2tqyY8eONx1j/fr13XWpAAAAHOC6LY7Hjx+fioqKjs/L5XJKpVKSpLq6Ojt37kxzc3Nq\na2s79qmurk5zc/M+239335qamn32/UPbf/8YAAAA8FYq/vAu74x+/f7/Dm9pacmgQYNSU1OTlpaW\nfbbX1tbus31/+w4aNCiVlZX7PcbAgQM79u2qpqamt3Op9HEH8+t7MF8bb5/1QWesDTpjbbA/1ged\nOVDXRo/F8XHHHZcNGzbkox/9aNauXZuTTz45w4cPz3e+853s2bMnra2teeqppzJs2LCMGjUqa9as\nyfDhw7N27dqMHj06NTU1qayszHPPPZcjjzwy69atywUXXJD+/fvnmmuuyVe+8pW89NJLaW9vz5Ah\nQzqOMWHChI5jdNW/Z98et2Jbb09wwOvTr+/b0NTUdNBeG2+f9UFnrA06Y22wP9YHnenra2N/4d5j\ncTx79uxcddVVaWxszNChQzN+/Pj0798/06ZNS0NDQ8rlcmbNmpWqqqpMnTo1s2fPztSpU1NZWZlF\nixYlSebPn59LLrkkbW1tqa+vz4gRI5IkY8aMyeTJk9Pe3p65c+cmSaZPn57Zs2dn9erVOfTQQzuO\nAQAAAL+vVC6Xy709RF/S1/9LxxkX39nbIxzw7l50Zm+P0C36+tqld1kfdMbaoDPWBvtjfdCZvr42\n9jdft/6dYwAAADgQiGMAAAAKTxwDAABQeOIYAACAwhPHAAAAFJ44BgAAoPDEMQAAAIUnjgEAACg8\ncQwAAEDhiWMAAAAKTxwDAABQeOIYAACAwhPHAAAAFJ44BgAAoPDEMQAAAIUnjgEAACg8cQwAAEDh\niWMAAAAKTxwDAABQeOIYAACAwhPHAAAAFJ44BgAAoPDEMQAAAIUnjgEAACg8cQwAAEDhiWMAAAAK\nTxwDAABQeOIYAACAwhPHAAAAFJ44BgAAoPDEMQAAAIUnjgEAACg8cQwAAEDhiWMAAAAKTxwDAABQ\neOIYAACAwhPHAAAAFJ44BgAAoPDEMQAAAIUnjgEAACg8cQwAAEDhiWMAAAAKTxwDAABQeOIYAACA\nwhPHAAAAFJ44BgAAoPDEMQAAAIUnjgEAACg8cQwAAEDhiWMAAAAKTxwDAABQeOIYAACAwqvoyZPt\n3bs3l112WV544YX069cv3/jGN1JRUZHLLrsspVIpRx99dObNm5d+/fpl9erVWbVqVSoqKjJ9+vSc\ncsop2b17dy699NK89tprqa6uzsKFCzNkyJBs3LgxV199dfr375/6+vpccMEFSZIlS5bkwQcfTEVF\nRebMmZPhw4f35OUCAABwgOjSneNzzz03P/7xj7N37963dbI1a9bkjTfeyKpVq3L++efnO9/5ThYs\nWJCZM2dmxYoVKZfLuf/++/PKK69k+fLlWbVqVZYtW5bGxsa0trZm5cqVGTZsWFasWJGzzjorS5cu\nTZLMmzcvixYtysqVK7Np06Y8/vjj2bJlSx599NHccsstaWxszPz589/W7AAAABy8uhTHf/VXf5WH\nHnoo48ePz/z587N58+Y/6mQf+tCH0tbWlvb29jQ3N6eioiJbtmzJSSedlCQZN25c1q9fn82bN2fk\nyJEZMGBAamtrU1dXl61bt6apqSljx47t2PeRRx5Jc3NzWltbU1dXl1KplPr6+qxfvz5NTU2pr69P\nqVTKEUcckba2tuzYseOPmhsAAICDW5ceqz7xxBNz4oknZvfu3bn33ntz4YUXpqamJhMnTkxDQ0MG\nDBjQpZO9613vygsvvJDTTz89r7/+eq677ro89thjKZVKSZLq6urs3Lkzzc3Nqa2t7fi+6urqNDc3\n77P9d/etqanZZ9/nn38+VVVVGTx48D7bd+7cmSFDhvzBOZuamrp0PRyYDubX92C+Nt4+64POWBt0\nxtpgf6wPOnOgro0u/87xhg0bcuedd+bhhx/OuHHj8hd/8Rd5+OGHM3369CxbtqxLx/if//N/pr6+\nPhdffHFefPHFfOlLX9rnUe2WlpYMGjQoNTU1aWlp2Wd7bW3tPtv3t++gQYNSWVn5lsfoitGjR3dp\nv16xYltvT3DA69Ov79vQ1NR00F4bb5/1QWesDTpjbbA/1ged6etrY3/h3qXHqk855ZQsWbIkJ510\nUn7yk5/kG9/4Rj72sY9l1qxZ/65HlQcNGtQRqO9+97vzxhtv5LjjjsuGDRuSJGvXrs2YMWMyfPjw\nNDU1Zc+ePdm5c2eeeuqpDBs2LKNGjcqaNWs69h09enRqampSWVmZ5557LuVyOevWrcuYMWMyatSo\nrFu3Lu3t7dm+fXva29u7dNcYAACA4unSnePvf//7qa6uznve857s3r07zz77bD74wQ+mf//+ueOO\nO7p8si9/+cuZM2dOGhoasnfv3syaNSvHH398rrrqqjQ2Nmbo0KEZP358+vfvn2nTpqWhoSHlcjmz\nZs1KVVVVpk6dmtmzZ2fq1KmprKzMokWLkiTz58/PJZdckra2ttTX12fEiBFJkjFjxmTy5Mlpb2/P\n3Llz/4gfDwAAAEXQpTh+8MEHc8cdd+SOO+7Ia6+9lvPOOy9f/vKXM3ny5H/Xyaqrq/O3f/u3b9p+\n0003vWnbpEmTMmnSpH22HXLIIbn22mvftO8JJ5yQ1atXv2n7jBkzMmPGjH/XjAAAABRPlx6rXr16\ndW6++eYkyfvf//7cfvvtbxm0AAAAcCDqUhzv3bt3n3ekrqys7LaBAAAAoKd16bHqU089NV/60pdy\n+umnJ0l++tOf5pOf/GS3DgYAAAA9pUtxfOmll+bee+/NY489loqKinzxi1/Mqaee2t2zAQAAQI/o\n8t85Puqoo3LYYYelXC4nSR577LGceOKJ3TYYAAAA9JQuxfH8+fPzwAMP5Mgjj+zYViqV8vd///fd\nNhgAAAD0lC7F8cMPP5x77703AwcO7O55AAAAoMd16d2qjzzyyI7HqQEAAOBg06U7x+9+97vzn//z\nf87IkSP3+ZNOCxYs6LbBAAAAoKd0KY7Hjh2bsWPHdvcsAAAA0Cu6FMef+9znsm3btjz55JOpr6/P\niy++uM+bcwEAAMCBrEu/c/yjH/0o06dPz9VXX51//dd/zZQpU3LnnXd292wAAADQI7oUxzfeeGNW\nrlyZ6urqvOc978kdd9yRG264obtnAwAAgB7RpTju169fampqOj4//PDD069fl74VAAAA+rwu/c7x\n0UcfnZtuuilvvPFGnnjiiaxYsSLHHHNMd88GAAAAPaJLt3/nzp2bl19+OVVVVZkzZ05qamoyb968\n7p4NAAAAekSX7hy/613vysUXX5yLL764u+cBAACAHtelOD7mmGNSKpX22fbe9743a9eu7ZahAAAA\noCd1KY63bt3a8fHevXtz3333ZePGjd02FAAAAPSkf/dbTldWVub000/P//7f/7s75gEAAIAe16U7\nxz/84Q87Pi6Xy/nlL3+ZysrKbhsKAAAAelKX4njDhg37fH7ooYfmb/7mb7plIAAAAOhpXYrjBQsW\ndPccAAAA0Gu6FMef/OQn3/Ru1cm/PWJdKpVy//33v+ODAQAAQE/pUhyfccYZqayszKRJk1JRUZG7\n7747P//5zzNr1qzung8AAAC6XZfi+KGHHsrtt9/e8fmXvvSlTJgwIe9///u7bTAAAADoKV3+U07r\n16/v+PiBBx5IdXV1twwEAAAAPa1Ld46//vWvZ/bs2Xn11VeTJEOHDs3ChQu7dTAAAADoKV2K4+OP\nPz733HNPduzYkaqqKneNAQAAOKh06bHqF154IX/5l3+ZKVOm5De/+U2++MUvZtu2bd09GwAAAPSI\nLsXx3Llz85WvfCXvete7cthhh+Uzn/lMZs+e3d2zAQAAQI/oUhy//vrrqa+vT5KUSqVMmjQpzc3N\n3ToYAAAA9JQuxfHAgQPz0ksvpVQqJUn+z//5PxkwYEC3DgYAAAA9pUtvyHX55Zfnv/7X/5rnnnsu\nZ555Zv71X/81f/u3f9vdswEAAECP6FIcv/baa7n11lvzzDPPpK2tLUOHDnXnGAAAgINGlx6rvuaa\na1JZWZmjjz46xxxzjDAGAADgoNKlO8dHHnlkLr/88owYMSIDBw7s2H7WWWd122AAAADQU/Ybxy+/\n/HL+5E/+JIceemiSZNOmTft8XRwDAABwMNhvHJ933nm54447smDBgnzve9/LOeec01NzAQAAQI/Z\n7+8cl8vljo/vvvvubh8GAAAAesN+4/i3f9c42TeUAQAA4GDSpXerTvYNZQAAADiY7Pd3jn/5y1/m\nU5/6VJJ/e3Ou335cLpdTKpVy//33d/+EAAAA0M32G8c/+clPemoOAAAA6DX7jeP3v//9PTUHAAAA\n9Jou/84xAAAAHKzEMQAAAIUnjgEAACg8cQwAAEDhiWMAAAAKTxwDAABQeOIYAACAwtvv3znuDtdf\nf31+9rOfZe/evZk6dWpOOumkXHbZZSmVSjn66KMzb9689OvXL6tXr86qVatSUVGR6dOn55RTTsnu\n3btz6aWX5rXXXkt1dXUWLlyYIUOGZOPGjbn66qvTv3//1NfX54ILLkiSLFmyJA8++GAqKioyZ86c\nDB8+vKcvFwAAgANAj9453rBhQ/7xH/8xK1euzPLly/PSSy9lwYIFmTlzZlasWJFyuZz7778/r7zy\nSpYvX55Vq1Zl2bJlaWxsTGtra1auXJlhw4ZlxYoVOeuss7J06dIkybx587Jo0aKsXLkymzZtyuOP\nP54tW7bk0UcfzS233JLGxsbMnz+/Jy8VAACAA0iPxvG6desybNiwnH/++TnvvPPyiU98Ilu2bMlJ\nJ52UJBk3blzWr1+fzZs3Z+TIkRkwYEBqa2tTV1eXrVu3pqmpKWPHju3Y95FHHklzc3NaW1tTV1eX\nUqmU+vr6rF+/Pk1NTamvr0+pVMoRRxyRtra27NixoycvFwAAgANEjz5W/frrr2f79u257rrrsm3b\ntkyfPj3lcjmlUilJUl1dnZ07d6a5uTm1tbUd31ddXZ3m5uZ9tv/uvjU1Nfvs+/zzz6eqqiqDBw/e\nZ/vOnTszZMiQPzhnU1PTO3XJ9EEH8+t7MF8bb5/1QWesDTpjbbA/1gedOVDXRo/G8eDBgzN06NAM\nGDAgQ4cOTVVVVV566aWOr7e0tGTQoEGpqalJS0vLPttra2v32b6/fQcNGpTKysq3PEZXjB49+u1e\navdZsa23Jzjg9enX921oamo6aK+Nt8/6oDPWBp2xNtgf64PO9PW1sb9w79HHqkePHp2HHnoo5XI5\nL7/8cnbt2pWPfexj2bBhQ5Jk7dq1GTNmTIYPH56mpqbs2bMnO3fuzFNPPZVhw4Zl1KhRWbNmTce+\no0ePTk1NTSorK/Pcc8+lXC5n3bp1GTNmTEaNGpV169alvb0927dvT3t7e5fuGgMAAFA8PXrn+JRT\nTsljjz2WiRMnplwuZ+7cufnABz6Qq666Ko2NjRk6dGjGjx+f/v37Z9q0aWloaEi5XM6sWbNSVVWV\nqVOnZvbs2Zk6dWoqKyuzaNGiJMn8+fNzySWXpK2tLfX19RkxYkSSZMyYMZk8eXLa29szd+7cnrxU\nAAAADiA9/qec/tt/+29v2nbTTTe9adukSZMyadKkfbYdcsghufbaa9+07wknnJDVq1e/afuMGTMy\nY8aMtzEtAAAARdCjj1UDAABAXySOAQAAKDxxDAAAQOGJYwAAAApPHAMAAFB44hgAAIDCE8cAAAAU\nnjgGAACg8MQxAAAAhSeOAQAAKDxxDAAAQOGJYwAAAApPHAMAAFB44hgAAIDCE8cAAAAUnjgGAACg\n8MQxAAAAhSeOAQAAKDxxDAAAQOGJYwAAAApPHAMAAFB44hgAAIDCE8cAAAAUnjgGAACg8MQxAAAA\nhSeOAQAAKDxxDAAAQOGJYwAAAApPHAMAAFB44hgAAIDCE8cAAAAUnjgGAACg8MQxAAAAhSeOAQAA\nKDxxDAAAQOGJYwAAAApPHAMAAFB44hgAAIDCE8cAAAAUnjgGAACg8MQxAAAAhSeOAQAAKDxxDAAA\nQOGJYwAAAApPHAMAAFB44hgAAIDCE8cAAAAUnjgGAACg8MQxAAAAhSeOAQAAKDxxDAAAQOH1Shy/\n9tpr+fjHP56nnnoqzz77bKZOnZqGhobMmzcv7e3tSZLVq1dnwoQJmTRpUh544IEkye7duzNjxow0\nNDTk3HPPzY4dO5IkGzduzNlnn50pU6ZkyZIlHedZsmRJJk6cmClTpmTz5s09f6EAAAAcEHo8jvfu\n3Zu5c+dm4MCBSZIFCxZk5syZWbFiRcrlcu6///688sorWb58eVatWpVly5alsbExra2tWblyZYYN\nG5YVK1bkrLPOytKlS5Mk8+bNy6JFi7Jy5cps2rQpjz/+eLZs2ZJHH300t9xySxobGzN//vyevlQA\nAAAOED0exwsXLsyUKVNy+OGHJ0m2bNmSk046KUkybty4rF+/Pps3b87IkSMzYMCA1NbWpq6uLlu3\nbk1TU1PGjh3bse8jjzyS5ubmtLa2pq6uLqVSKfX19Vm/fn2amppSX1+fUqmUI444Im1tbR13mgEA\nAOB3VfTkyW6//fYMGTIkY8eOzQ033JAkKZfLKZVKSZLq6urs3Lkzzc3Nqa2t7fi+6urqNDc377P9\nd/etqanZZ9/nn38+VVVVGTx48D7bd+7cmSFDhvzBOZuamt6R66VvOphf34P52nj7rA86Y23QGWuD\n/bE+6MyVIDVoAAAUO0lEQVSBujZ6NI5vu+22lEqlPPLII3niiScye/bsfe7mtrS0ZNCgQampqUlL\nS8s+22tra/fZvr99Bw0alMrKyrc8RleMHj367V5q91mxrbcnOOD16df3bWhqajpor423z/qgM9YG\nnbE22B/rg8709bWxv3Dv0ceqb7755tx0001Zvnx5jj322CxcuDDjxo3Lhg0bkiRr167NmDFjMnz4\n8DQ1NWXPnj3ZuXNnnnrqqQwbNiyjRo3KmjVrOvYdPXp0ampqUllZmeeeey7lcjnr1q3LmDFjMmrU\nqKxbty7t7e3Zvn172tvbu3TXGAAAgOLp0TvHb2X27Nm56qqr0tjYmKFDh2b8+PHp379/pk2bloaG\nhpTL5cyaNStVVVWZOnVqZs+enalTp6aysjKLFi1KksyfPz+XXHJJ2traUl9fnxEjRiRJxowZk8mT\nJ6e9vT1z587tzcsEAACgD+u1OF6+fHnHxzfddNObvj5p0qRMmjRpn22HHHJIrr322jfte8IJJ2T1\n6tVv2j5jxozMmDHjHZgWAACAg1mv/J1jAAAA6EvEMQAAAIUnjgEAACg8cQwAAEDhiWMAAAAKTxwD\nAABQeOIYAACAwhPHAAAAFJ44BgAAoPDEMQAAAIUnjgEAACg8cQwAAEDhiWMAAAAKTxwDAABQeOIY\nAACAwhPHAAAAFJ44BgAAoPDEMQAAAIUnjgEAACg8cQwAAEDhiWMAAAAKTxwDAABQeOIYAACAwhPH\nAAAAFJ44BgAAoPDEMQAAAIVX0dsDQE874+I7e3uE7rNiW4+c5u5FZ/bIeQAAoKe4cwwAAEDhiWMA\nAAAKTxwDAABQeOIYAACAwhPHAAAAFJ44BgAAoPDEMQAAAIUnjgEAACg8cQwAAEDhiWMAAAAKTxwD\nAABQeOIYAACAwhPHAAAAFJ44BgAAoPDEMQAAAIUnjgEAACg8cQwAAEDhiWMAAAAKTxwDAABQeOIY\nAACAwhPHAAAAFJ44BgAAoPDEMQAAAIUnjgEAACg8cQwAAEDhVfTkyfbu3Zs5c+bkhRdeSGtra6ZP\nn54Pf/jDueyyy1IqlXL00Udn3rx56devX1avXp1Vq1aloqIi06dPzymnnJLdu3fn0ksvzWuvvZbq\n6uosXLgwQ4YMycaNG3P11Venf//+qa+vzwUXXJAkWbJkSR588MFUVFRkzpw5GT58eE9eLgAAAAeI\nHo3ju+66K4MHD84111yTX/3qVznrrLNyzDHHZObMmfnoRz+auXPn5v77788JJ5yQ5cuX57bbbsue\nPXvS0NCQP//zP8/KlSszbNiwzJgxI/fcc0+WLl2aK6+8MvPmzcvixYtz5JFH5q/+6q/y+OOPp1wu\n59FHH80tt9ySF198MTNmzMhtt93Wk5cLAADAAaJH4/i0007L+PHjkyTlcjn9+/fPli1bctJJJyVJ\nxo0bl4cffjj9+vXLyJEjM2DAgAwYMCB1dXXZunVrmpqa8tWvfrVj36VLl6a5uTmtra2pq6tLktTX\n12f9+vUZMGBA6uvrUyqVcsQRR6StrS07duzIkCFDevKSAQAAOAD0aBxXV1cnSZqbm3PhhRdm5syZ\nWbhwYUqlUsfXd+7cmebm5tTW1u7zfc3Nzfts/919a2pq9tn3+eefT1VVVQYPHrzP9p07d3Ypjpua\nmt6R64WDlX9GDkxeNzpjbdAZa4P9sT7ozIG6Nno0jpPkxRdfzPnnn5+GhoacccYZueaaazq+1tLS\nkkGDBqWmpiYtLS37bK+trd1n+/72HTRoUCorK9/yGF0xevTot3uZ3WfFtt6eAPr2PyO8paamJq8b\nb8naoDPWBvtjfdCZvr429hfuPfpu1a+++mrOOeecXHrppZk4cWKS5LjjjsuGDRuSJGvXrs2YMWMy\nfPjwNDU1Zc+ePdm5c2eeeuqpDBs2LKNGjcqaNWs69h09enRqampSWVmZ5557LuVyOevWrcuYMWMy\natSorFu3Lu3t7dm+fXva29s9Ug0AAMBb6tE7x9ddd11+/etfZ+nSpVm6dGmS5Iorrsg3v/nNNDY2\nZujQoRk/fnz69++fadOmpaGhIeVyObNmzUpVVVWmTp2a2bNnZ+rUqamsrMyiRYuSJPPnz88ll1yS\ntra21NfXZ8SIEUmSMWPGZPLkyWlvb8/cuXN78lIBAAA4gJTK5XK5t4foS/r6YwBnXHxnb48AuXvR\nmb09Av9Off3fbfQea4POWBvsj/VBZ/r62tjffD36WDUAAAD0ReIYAACAwhPHAAAAFJ44BgAAoPDE\nMQAAAIUnjgEAACg8cQwAAEDhiWMAAAAKTxwDAABQeOIYAACAwhPHAAAAFJ44BgAAoPAqensAAIAD\n2RkX39nbI3SfFdt65DR3LzqzR84DsD/uHAMAAFB44hgAAIDCE8cAAAAUnjgGAACg8MQxAAAAhSeO\nAQAAKDxxDAAAQOGJYwAAAApPHAMAAFB44hgAAIDCE8cAAAAUnjgGAACg8MQxAAAAhSeOAQAAKDxx\nDAAAQOFV9PYAAEV1xsV39uwJV2zr2fP1gLsXndnbIwAABwl3jgEAACg8cQwAAEDhiWMAAAAKTxwD\nAABQeOIYAACAwhPHAAAAFJ44BgAAoPDEMQAAAIUnjgEAACg8cQwAAEDhiWMAAAAKTxwDAABQeOIY\nAACAwqvo7QEAgN7ztRXbkhXbensM4G064+I7e/6kB+G/O+5edGZvj0AvcucYAACAwhPHAAAAFJ44\nBgAAoPDEMQAAAIUnjgEAACg8cQwAAEDhiWMAAAAKTxwDAABQeOIYAACAwqvo7QG6U3t7e772ta/l\nF7/4RQYMGJBvfvOb+eAHP9jbYwEAANDHHNR3ju+77760trbmBz/4QS6++OJ8+9vf7u2RAAAA6IMO\n6jhuamrK2LFjkyQnnHBC/umf/qmXJwIAAKAvKpXL5XJvD9Fdrrjiinz605/Oxz/+8STJJz7xidx3\n332pqOj8afKmpqaeGg8AAIAeNnr06LfcflD/znFNTU1aWlo6Pm9vb99vGCed/6AAAAA4eB3Uj1WP\nGjUqa9euTZJs3Lgxw4YN6+WJAAAA6IsO6seqf/tu1f/8z/+ccrmcb33rWznqqKN6eywAAAD6mIM6\njgEAAKArDurHqgEAAKArxDEAAACFd1C/W/XBZtOmTfnrv/7rLF++vLdHoQ/Zu3dv5syZkxdeeCGt\nra2ZPn16PvWpT/X2WPQBbW1tufLKK/P000+nVCpl/vz53piQfbz22muZMGFCvve973lPDvbxuc99\nLjU1NUmSD3zgA1mwYEEvT0Rfcf311+dnP/tZ9u7dm6lTp+bss8/u7ZHoA26//fbccccdSZI9e/bk\niSeeyMMPP5xBgwb18mT/PuL4AHHjjTfmrrvuyiGHHNLbo9DH3HXXXRk8eHCuueaa/OpXv8pZZ50l\njkmSPPDAA0mSVatWZcOGDfmbv/mbfPe73+3lqegr9u7dm7lz52bgwIG9PQp9zJ49e1Iul/3HeN5k\nw4YN+cd//MesXLkyu3btyve+973eHok+YsKECZkwYUKSZP78+fn85z9/wIVx4rHqA0ZdXV0WL17c\n22PQB5122mm56KKLkiTlcjn9+/fv5YnoK0499dR84xvfSJJs3779gPwfKbrPwoULM2XKlBx++OG9\nPQp9zNatW7Nr166cc845+eIXv5iNGzf29kj0EevWrcuwYcNy/vnn57zzzssnPvGJ3h6JPubnP/95\nnnzyyUyePLm3R/mjuHN8gBg/fny2bdvW22PQB1VXVydJmpubc+GFF2bmzJm9PBF9SUVFRWbPnp1/\n+Id/yLXXXtvb49BH3H777RkyZEjGjh2bG264obfHoY8ZOHBgvvKVr+Tss8/OM888k3PPPTf33ntv\nKir838aie/3117N9+/Zcd9112bZtW6ZPn5577703pVKpt0ejj7j++utz/vnn9/YYfzR3juEg8OKL\nL+aLX/xizjzzzJxxxhm9PQ59zMKFC/OTn/wkV111VX7zm9/09jj0AbfddlvWr1+fadOm5Yknnsjs\n2bPzyiuv9PZY9BEf+tCH8tnPfjalUikf+tCHMnjwYOuDJMngwYNTX1+fAQMGZOjQoamqqsqOHTt6\neyz6iF//+td5+umnc/LJJ/f2KH80cQwHuFdffTXnnHNOLr300kycOLG3x6EP+eEPf5jrr78+SXLI\nIYekVCqlXz//2ie5+eabc9NNN2X58uU59thjs3Dhwrz3ve/t7bHoI2699dZ8+9vfTpK8/PLLaW5u\ntj5IkowePToPPfRQyuVyXn755ezatSuDBw/u7bHoIx577LF87GMf6+0x3hbPx8AB7rrrrsuvf/3r\nLF26NEuXLk3yb2/g5k12+PSnP53LL788X/jCF/LGG29kzpw51gXwB02cODGXX355pk6dmlKplG99\n61seqSZJcsopp+Sxxx7LxIkTUy6XM3fuXO91Qoenn346H/jAB3p7jLelVC6Xy709BAAAAPQmz9cB\nAABQeOIYAACAwhPHAAAAFJ44BgAAoPDEMQAAAIXnffkB4B22bdu2nHbaaTnqqKNSKpWyd+/eHH74\n4VmwYEH+9E//tLfHe0c8+OCDue666/Kb3/wm7e3tOfXUU3PhhRe+o39L+/nnn893v/vdfOtb33rH\njgkAnXHnGAC6weGHH54777wzP/zhD3PPPffk+OOPzze+8Y3eHusdsXbt2nz961/PggULctddd+XW\nW2/N1q1bc+21176j59m+fXuef/75d/SYANAZd44BoAeMGTMmP/vZz5IkP/7xj/N3f/d32b17d/bs\n2ZNvfvObOfHEE/N3f/d3ueOOO9KvX78MHz48X//617N169bMnTs3b7zxRqqqqrJgwYL8h//wH3LT\nTTflzjvvzK5du1IqlfKd73wnRx11VD75yU/ms5/9bNatW5ddu3Zl4cKFOf744/PP//zPueyyy9LW\n1pYxY8Zk7dq1+Yd/+Ie8+uqrmTt3bl566aWUSqVcfPHF+bM/+7MsXrw4GzduzIsvvpgvfOEL+cIX\nvtBxLdddd10uuOCCfOhDH0qSDBw4MF/72tfyL//yL0mSp59+OnPnzs2vfvWrvOtd78oVV1yR4cOH\n57LLLstJJ52UCRMmJEk+8pGP5Be/+EUWL16cl19+Oc8++2xeeOGFnH322Zk+fXq++c1vZtu2bZk/\nf37mzZvXw68YAEXjzjEAdLO9e/fmxz/+cUaNGpX29vasWrUq1113Xe66666ce+65WbZsWd54441c\nf/31ue2223L77benVCrl5Zdfzve///385V/+ZW6//fZMmzYtGzduTHNzc+67774sX748/+t//a+c\neuqpWbFiRcf5Bg8enFtvvTVTpkzJ9ddfnyS57LLLctFFF+XOO+/MkUcemba2tiTJ1Vdfnc9//vO5\n/fbb893vfjdz585Nc3NzkqS1tTU/+tGP9gnjJHniiScyYsSIfbb96Z/+af7sz/4sSXLppZdm2rRp\nufvuu3P55ZfnoosuSmtr635/Rr/4xS+ybNmy3HLLLbnhhhvy61//OldeeWWOP/54YQxAj3DnGAC6\nwf/7f/8vZ555ZpJ/i8zhw4fn4osvTr9+/fI//sf/yM9+9rM8/fTTefTRR9OvX79UVFRk5MiRmThx\nYj71qU/lC1/4Qv7kT/4kH//4x/P1r389Dz30UE455ZSMHz8+/fv3z6JFi3LPPffkmWeeyUMPPZRj\njz2249xjx45Nkhx99NH56U9/ml/96ld54YUX8vGPfzxJ8vnPfz5///d/nyRZv359/uVf/qXjkeg3\n3nij41Hm4cOHv+W1lUqllMvlt/xaS0tLnnvuuXz6059Okpxwwgl597vf3XFXuTMf/ehHM2DAgLzn\nPe/J4MGDs3Pnzi79nAHgnSKOAaAb/PZ3jn9fS0tLPv/5z+fMM8/MiSeemI985CO5+eabkyRLly7N\nxo0bs3bt2nz1q1/NX//1X+e0007LyJEj88ADD+T73/9+1qxZk/PPPz/Tpk3Lf/kv/yXjxo3LYYcd\nlieeeKLjHFVVVUn+LWKTpH///p3GbHt7e77//e9n8ODBSZKXX345hx12WO67774MHDjwLb/n+OOP\nzz/90z/lwx/+cMe2p59+uuPO8++fq1wup62tbZ+o3rt37z77/Hbm387d2bwA0F08Vg0APeiZZ55J\nv379ct555+Xkk0/O2rVr09bWlh07duT000/PsGHDctFFF+XP//zP84tf/CIzZ87M5s2bM2XKlFx0\n0UV5/PHH8/Of/zwf/OAH8+UvfzkjRozoOEZnamtrU1dXlzVr1iRJ7r777o6vnXzyyR2PZD/55JP5\n7Gc/m127du33Gr761a9myZIleeaZZ5L8W/B/+9vfzvve977U1NTkyCOPzE9/+tMkycaNG/Pqq6/m\n6KOPzuDBg/Pkk08mSe67774/+LPq379/3njjjT+4HwC8E9w5BoAedMwxx+TYY4/N6aefnoEDB+bE\nE0/M9u3bM2TIkEyZMiUTJ07MIYcckve973353Oc+lxNPPDFXXHFFli5dmv79++eyyy7Lf/yP/zEr\nV67MX/zFX2TAgAEZPnx4fvnLX+73vAsXLsycOXPyne98Jx/5yEc67gpfeeWVmTt3bs4444wkyX//\n7/89NTU1+z3WuHHjMmvWrMyaNSttbW154403ctppp+WCCy5IklxzzTX52te+lsWLF6eysjKLFy/O\ngAED0tDQkJkzZ+aMM87IySefnPe+9737Pc9RRx2VnTt35tJLL80111zT1R8xAPxRSmXPLQHAQW/J\nkiWZNGlSDj/88Pz0pz/N3XffncWLF/f2WADQZ7hzDAAFcMQRR+Scc85JRUVFBg0alKuvvrq3RwKA\nPsWdYwDg/2vnjgkAAAAAgvVvLYithQcA7BlyAQAAsCeOAQAA2BPHAAAA7IljAAAA9sQxAAAAe+IY\nAACAvQD3PVmlH7D06QAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x39d26e9d0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "train_sample.passenger_count.hist(bins=10, figsize = (16,8))\n",
    "plt.xlabel(\"Passanger Count\")\n",
    "plt.ylabel(\"Frequency\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x39cf32110>"
      ]
     },
     "execution_count": 41,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAA7EAAAHfCAYAAACGQl9aAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nFgAAgMT4/wGZFCttDeN77AAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x39ce54790>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize= (16,8))\n",
    "sns.boxplot(x = train_sample.passenger_count, y = train.fare_amount)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAngAAAFKCAYAAACOzSIBAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x3a016c350>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "f, ax = plt.subplots(1,2,figsize = [10,5])\n",
    "sns.countplot(train_sample[\"passenger_count\"], ax=ax[0])\n",
    "sns.countplot(test[\"passenger_count\"], ax=ax[1])\n",
    "ax[0].set_title(\"Train Set - Passenger Count\")\n",
    "ax[1].set_title(\"Test Set - Passenger Count\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# feature engineering\n",
    "# time\n",
    "# distance"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def prepare_time_features(df):\n",
    "    df['pickup_datetime'] = df['pickup_datetime'].str.replace(\" UTC\", \"\")\n",
    "    df['pickup_datetime'] = pd.to_datetime(df['pickup_datetime'], format='%Y-%m-%d %H:%M:%S')\n",
    "    df['hour_of_day'] = df.pickup_datetime.dt.hour\n",
    "    df['week'] = df.pickup_datetime.dt.week\n",
    "    df['month'] = df.pickup_datetime.dt.month\n",
    "    df[\"year\"] = df.pickup_datetime.dt.year\n",
    "    df['day_of_year'] = df.pickup_datetime.dt.dayofyear\n",
    "    df['week_of_year'] = df.pickup_datetime.dt.weekofyear\n",
    "    df[\"weekday\"] = df.pickup_datetime.dt.weekday\n",
    "    df[\"quarter\"] = df.pickup_datetime.dt.quarter\n",
    "    df[\"day_of_month\"] = df.pickup_datetime.dt.day\n",
    "    return df"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "train = prepare_time_features(train)\n",
    "test = prepare_time_features(test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "train.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(9914, 16)"
      ]
     },
     "execution_count": 45,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "test.shape # add 9 time features"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# calculate-distance-between-two-latitude-longitude-points-haversine-formula \n",
    "# Returns distance in miles\n",
    "def distance(lat1, lon1, lat2, lon2):\n",
    "    p = 0.017453292519943295 # Pi/180\n",
    "    a = 0.5 - np.cos((lat2 - lat1) * p)/2 + np.cos(lat1 * p) * np.cos(lat2 * p) * (1 - np.cos((lon2 - lon1) * p)) / 2\n",
    "    return 0.6213712 * 12742 * np.arcsin(np.sqrt(a))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "train['distance_miles'] = distance(train.pickup_latitude, train.pickup_longitude, \\\n",
    "                                      train.dropoff_latitude, train.dropoff_longitude)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "test['distance_miles'] = distance(test.pickup_latitude, test.pickup_longitude, \\\n",
    "                                      test.dropoff_latitude, test.dropoff_longitude)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(178, 18)"
      ]
     },
     "execution_count": 49,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train[(train['distance_miles']==0)&(train['fare_amount']==0)].shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(54055716, 18)"
      ]
     },
     "execution_count": 51,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train = train.drop(train[(train['distance_miles']==0)&(train['fare_amount']==0)].index, axis=0)\n",
    "train.shape # drop 178 obs "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Calculating pickup and drop distance from all 3 airports of Air Ports\n",
    "def transform(data):\n",
    "    # Distances to nearby airports, \n",
    "    jfk = (-73.7781, 40.6413)\n",
    "    ewr = (-74.1745, 40.6895)\n",
    "    lgr = (-73.8740, 40.7769)\n",
    "\n",
    "    data['pickup_distance_to_jfk'] = distance(jfk[1], jfk[0],\n",
    "                                         data['pickup_latitude'], data['pickup_longitude'])\n",
    "    data['dropoff_distance_to_jfk'] = distance(jfk[1], jfk[0],\n",
    "                                           data['dropoff_latitude'], data['dropoff_longitude'])\n",
    "    data['pickup_distance_to_ewr'] = distance(ewr[1], ewr[0], \n",
    "                                          data['pickup_latitude'], data['pickup_longitude'])\n",
    "    data['dropoff_distance_to_ewr'] = distance(ewr[1], ewr[0],\n",
    "                                           data['dropoff_latitude'], data['dropoff_longitude'])\n",
    "    data['pickup_distance_to_lgr'] = distance(lgr[1], lgr[0],\n",
    "                                          data['pickup_latitude'], data['pickup_longitude'])\n",
    "    data['dropoff_distance_to_lgr'] = distance(lgr[1], lgr[0],\n",
    "                                           data['dropoff_latitude'], data['dropoff_longitude'])\n",
    "    \n",
    "    return data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "train = transform(train)\n",
    "test = transform(test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(54055716, 24)"
      ]
     },
     "execution_count": 54,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train.shape # add 7 distance features totally"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(1651, 24)"
      ]
     },
     "execution_count": 55,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train[train['fare_amount'] < 2.5].shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "#the base fare for any taxi in new york is 2.5 dollars, we will drop those cases"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "train = train.drop(train[train['fare_amount'] < 2.5].index, axis=0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(54054065, 24)"
      ]
     },
     "execution_count": 58,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 59,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(9914, 23)"
      ]
     },
     "execution_count": 59,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "test.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 61,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>key</th>\n",
       "      <th>fare_amount</th>\n",
       "      <th>pickup_datetime</th>\n",
       "      <th>pickup_longitude</th>\n",
       "      <th>pickup_latitude</th>\n",
       "      <th>dropoff_longitude</th>\n",
       "      <th>dropoff_latitude</th>\n",
       "      <th>passenger_count</th>\n",
       "      <th>hour_of_day</th>\n",
       "      <th>week</th>\n",
       "      <th>...</th>\n",
       "      <th>weekday</th>\n",
       "      <th>quarter</th>\n",
       "      <th>day_of_month</th>\n",
       "      <th>distance_miles</th>\n",
       "      <th>pickup_distance_to_jfk</th>\n",
       "      <th>dropoff_distance_to_jfk</th>\n",
       "      <th>pickup_distance_to_ewr</th>\n",
       "      <th>dropoff_distance_to_ewr</th>\n",
       "      <th>pickup_distance_to_lgr</th>\n",
       "      <th>dropoff_distance_to_lgr</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>2009-06-15 17:26:21.0000001</td>\n",
       "      <td>4.5</td>\n",
       "      <td>2009-06-15 17:26:21</td>\n",
       "      <td>-73.844311</td>\n",
       "      <td>40.721319</td>\n",
       "      <td>-73.841610</td>\n",
       "      <td>40.712278</td>\n",
       "      <td>1</td>\n",
       "      <td>17</td>\n",
       "      <td>25</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>15</td>\n",
       "      <td>0.640487</td>\n",
       "      <td>6.527098</td>\n",
       "      <td>5.926672</td>\n",
       "      <td>17.433727</td>\n",
       "      <td>17.508086</td>\n",
       "      <td>4.142790</td>\n",
       "      <td>4.776039</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2010-01-05 16:52:16.0000002</td>\n",
       "      <td>16.9</td>\n",
       "      <td>2010-01-05 16:52:16</td>\n",
       "      <td>-74.016048</td>\n",
       "      <td>40.711303</td>\n",
       "      <td>-73.979268</td>\n",
       "      <td>40.782004</td>\n",
       "      <td>1</td>\n",
       "      <td>16</td>\n",
       "      <td>1</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>5</td>\n",
       "      <td>5.250670</td>\n",
       "      <td>13.373879</td>\n",
       "      <td>14.335733</td>\n",
       "      <td>8.435586</td>\n",
       "      <td>12.054959</td>\n",
       "      <td>8.708241</td>\n",
       "      <td>5.518847</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2011-08-18 00:35:00.00000049</td>\n",
       "      <td>5.7</td>\n",
       "      <td>2011-08-18 00:35:00</td>\n",
       "      <td>-73.982738</td>\n",
       "      <td>40.761270</td>\n",
       "      <td>-73.991242</td>\n",
       "      <td>40.750562</td>\n",
       "      <td>2</td>\n",
       "      <td>0</td>\n",
       "      <td>33</td>\n",
       "      <td>...</td>\n",
       "      <td>3</td>\n",
       "      <td>3</td>\n",
       "      <td>18</td>\n",
       "      <td>0.863411</td>\n",
       "      <td>13.550258</td>\n",
       "      <td>13.478114</td>\n",
       "      <td>11.198778</td>\n",
       "      <td>10.482997</td>\n",
       "      <td>5.791580</td>\n",
       "      <td>6.399678</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>2012-04-21 04:30:42.0000001</td>\n",
       "      <td>7.7</td>\n",
       "      <td>2012-04-21 04:30:42</td>\n",
       "      <td>-73.987130</td>\n",
       "      <td>40.733143</td>\n",
       "      <td>-73.991567</td>\n",
       "      <td>40.758092</td>\n",
       "      <td>1</td>\n",
       "      <td>4</td>\n",
       "      <td>16</td>\n",
       "      <td>...</td>\n",
       "      <td>5</td>\n",
       "      <td>2</td>\n",
       "      <td>21</td>\n",
       "      <td>1.739386</td>\n",
       "      <td>12.657169</td>\n",
       "      <td>13.789559</td>\n",
       "      <td>10.266000</td>\n",
       "      <td>10.687245</td>\n",
       "      <td>6.648280</td>\n",
       "      <td>6.287901</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>2010-03-09 07:51:00.000000135</td>\n",
       "      <td>5.3</td>\n",
       "      <td>2010-03-09 07:51:00</td>\n",
       "      <td>-73.968095</td>\n",
       "      <td>40.768008</td>\n",
       "      <td>-73.956655</td>\n",
       "      <td>40.783762</td>\n",
       "      <td>1</td>\n",
       "      <td>7</td>\n",
       "      <td>10</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>9</td>\n",
       "      <td>1.242218</td>\n",
       "      <td>13.254398</td>\n",
       "      <td>13.577000</td>\n",
       "      <td>12.092157</td>\n",
       "      <td>13.133498</td>\n",
       "      <td>4.961705</td>\n",
       "      <td>4.350323</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 24 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "                             key  fare_amount     pickup_datetime  \\\n",
       "0    2009-06-15 17:26:21.0000001          4.5 2009-06-15 17:26:21   \n",
       "1    2010-01-05 16:52:16.0000002         16.9 2010-01-05 16:52:16   \n",
       "2   2011-08-18 00:35:00.00000049          5.7 2011-08-18 00:35:00   \n",
       "3    2012-04-21 04:30:42.0000001          7.7 2012-04-21 04:30:42   \n",
       "4  2010-03-09 07:51:00.000000135          5.3 2010-03-09 07:51:00   \n",
       "\n",
       "   pickup_longitude  pickup_latitude  dropoff_longitude  dropoff_latitude  \\\n",
       "0        -73.844311        40.721319         -73.841610         40.712278   \n",
       "1        -74.016048        40.711303         -73.979268         40.782004   \n",
       "2        -73.982738        40.761270         -73.991242         40.750562   \n",
       "3        -73.987130        40.733143         -73.991567         40.758092   \n",
       "4        -73.968095        40.768008         -73.956655         40.783762   \n",
       "\n",
       "   passenger_count  hour_of_day  week           ...             weekday  \\\n",
       "0                1           17    25           ...                   0   \n",
       "1                1           16     1           ...                   1   \n",
       "2                2            0    33           ...                   3   \n",
       "3                1            4    16           ...                   5   \n",
       "4                1            7    10           ...                   1   \n",
       "\n",
       "   quarter  day_of_month  distance_miles  pickup_distance_to_jfk  \\\n",
       "0        2            15        0.640487                6.527098   \n",
       "1        1             5        5.250670               13.373879   \n",
       "2        3            18        0.863411               13.550258   \n",
       "3        2            21        1.739386               12.657169   \n",
       "4        1             9        1.242218               13.254398   \n",
       "\n",
       "   dropoff_distance_to_jfk  pickup_distance_to_ewr  dropoff_distance_to_ewr  \\\n",
       "0                 5.926672               17.433727                17.508086   \n",
       "1                14.335733                8.435586                12.054959   \n",
       "2                13.478114               11.198778                10.482997   \n",
       "3                13.789559               10.266000                10.687245   \n",
       "4                13.577000               12.092157                13.133498   \n",
       "\n",
       "   pickup_distance_to_lgr  dropoff_distance_to_lgr  \n",
       "0                4.142790                 4.776039  \n",
       "1                8.708241                 5.518847  \n",
       "2                5.791580                 6.399678  \n",
       "3                6.648280                 6.287901  \n",
       "4                4.961705                 4.350323  \n",
       "\n",
       "[5 rows x 24 columns]"
      ]
     },
     "execution_count": 61,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 62,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "train.to_csv('train_model.csv', index=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 63,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "test.to_csv('test_model.csv', index=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 64,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# model\n",
    "# lightgbm\n",
    "# xgboost\n",
    "# random forest\n",
    "# catboost"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "import lightgbm as lgb\n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn import metrics\n",
    "from sklearn.metrics import mean_squared_error\n",
    "from sklearn.model_selection import KFold\n",
    "import gc"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# two big dataset: sample 5,000,000"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "train=pd.read_csv('train_model.csv')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "#train= train.sample(n=5000000)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(5000000, 24)"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "test=pd.read_csv('test_model.csv')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "feats = [f for f in train.columns if f not in ['key','pickup_datetime','fare_amount']]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training until validation scores don't improve for 125 rounds.\n",
      "[500]\ttrain's rmse: 3.67372\tvalid's rmse: 3.7396\n",
      "[1000]\ttrain's rmse: 3.57068\tvalid's rmse: 3.66357\n",
      "[1500]\ttrain's rmse: 3.5137\tvalid's rmse: 3.62972\n",
      "[2000]\ttrain's rmse: 3.4722\tvalid's rmse: 3.61049\n",
      "[2500]\ttrain's rmse: 3.43951\tvalid's rmse: 3.59745\n",
      "[3000]\ttrain's rmse: 3.41184\tvalid's rmse: 3.58789\n",
      "[3500]\ttrain's rmse: 3.38802\tvalid's rmse: 3.58103\n",
      "[4000]\ttrain's rmse: 3.36618\tvalid's rmse: 3.57597\n",
      "[4500]\ttrain's rmse: 3.34638\tvalid's rmse: 3.57188\n",
      "[5000]\ttrain's rmse: 3.32722\tvalid's rmse: 3.56797\n",
      "[5500]\ttrain's rmse: 3.3105\tvalid's rmse: 3.56538\n",
      "[6000]\ttrain's rmse: 3.29462\tvalid's rmse: 3.56322\n",
      "[6500]\ttrain's rmse: 3.27976\tvalid's rmse: 3.56179\n",
      "Early stopping, best iteration is:\n",
      "[6431]\ttrain's rmse: 3.28178\tvalid's rmse: 3.56173\n",
      "Fold  1 rmse : 3.561726\n",
      "Training until validation scores don't improve for 125 rounds.\n",
      "[500]\ttrain's rmse: 3.66582\tvalid's rmse: 3.7858\n",
      "[1000]\ttrain's rmse: 3.56642\tvalid's rmse: 3.71439\n",
      "[1500]\ttrain's rmse: 3.50757\tvalid's rmse: 3.67982\n",
      "[2000]\ttrain's rmse: 3.46663\tvalid's rmse: 3.66075\n",
      "[2500]\ttrain's rmse: 3.4356\tvalid's rmse: 3.6484\n",
      "[3000]\ttrain's rmse: 3.40822\tvalid's rmse: 3.63937\n",
      "[3500]\ttrain's rmse: 3.38477\tvalid's rmse: 3.63344\n",
      "[4000]\ttrain's rmse: 3.36418\tvalid's rmse: 3.62913\n",
      "[4500]\ttrain's rmse: 3.34621\tvalid's rmse: 3.6246\n",
      "[5000]\ttrain's rmse: 3.32927\tvalid's rmse: 3.62086\n",
      "[5500]\ttrain's rmse: 3.31354\tvalid's rmse: 3.61811\n",
      "[6000]\ttrain's rmse: 3.29936\tvalid's rmse: 3.61612\n",
      "Early stopping, best iteration is:\n",
      "[6278]\ttrain's rmse: 3.29159\tvalid's rmse: 3.61523\n",
      "Fold  2 rmse : 3.615227\n",
      "Training until validation scores don't improve for 125 rounds.\n",
      "[500]\ttrain's rmse: 3.68669\tvalid's rmse: 3.68323\n",
      "[1000]\ttrain's rmse: 3.58631\tvalid's rmse: 3.60712\n",
      "[1500]\ttrain's rmse: 3.5297\tvalid's rmse: 3.57269\n",
      "[2000]\ttrain's rmse: 3.48908\tvalid's rmse: 3.55241\n",
      "[2500]\ttrain's rmse: 3.45628\tvalid's rmse: 3.53952\n",
      "[3000]\ttrain's rmse: 3.42659\tvalid's rmse: 3.52813\n",
      "[3500]\ttrain's rmse: 3.4029\tvalid's rmse: 3.52242\n",
      "[4000]\ttrain's rmse: 3.38074\tvalid's rmse: 3.51604\n",
      "[4500]\ttrain's rmse: 3.36166\tvalid's rmse: 3.51215\n",
      "[5000]\ttrain's rmse: 3.34443\tvalid's rmse: 3.50846\n",
      "[5500]\ttrain's rmse: 3.32876\tvalid's rmse: 3.50597\n",
      "[6000]\ttrain's rmse: 3.31214\tvalid's rmse: 3.50332\n",
      "[6500]\ttrain's rmse: 3.2957\tvalid's rmse: 3.50066\n",
      "Early stopping, best iteration is:\n",
      "[6620]\ttrain's rmse: 3.29246\tvalid's rmse: 3.50021\n",
      "Fold  3 rmse : 3.500209\n",
      "Training until validation scores don't improve for 125 rounds.\n",
      "[500]\ttrain's rmse: 3.68879\tvalid's rmse: 3.67235\n",
      "[1000]\ttrain's rmse: 3.58745\tvalid's rmse: 3.60041\n",
      "[1500]\ttrain's rmse: 3.52932\tvalid's rmse: 3.56932\n",
      "[2000]\ttrain's rmse: 3.48573\tvalid's rmse: 3.54986\n",
      "[2500]\ttrain's rmse: 3.45253\tvalid's rmse: 3.5378\n",
      "[3000]\ttrain's rmse: 3.42394\tvalid's rmse: 3.53048\n",
      "[3500]\ttrain's rmse: 3.39907\tvalid's rmse: 3.52467\n",
      "[4000]\ttrain's rmse: 3.37761\tvalid's rmse: 3.52073\n",
      "[4500]\ttrain's rmse: 3.35803\tvalid's rmse: 3.51702\n",
      "[5000]\ttrain's rmse: 3.3402\tvalid's rmse: 3.51329\n",
      "[5500]\ttrain's rmse: 3.32485\tvalid's rmse: 3.51132\n",
      "[6000]\ttrain's rmse: 3.31011\tvalid's rmse: 3.50864\n",
      "[6500]\ttrain's rmse: 3.29579\tvalid's rmse: 3.50724\n",
      "[7000]\ttrain's rmse: 3.28196\tvalid's rmse: 3.50564\n",
      "Early stopping, best iteration is:\n",
      "[7160]\ttrain's rmse: 3.27776\tvalid's rmse: 3.50525\n",
      "Fold  4 rmse : 3.505248\n",
      "Training until validation scores don't improve for 125 rounds.\n",
      "[500]\ttrain's rmse: 3.68187\tvalid's rmse: 3.69773\n",
      "[1000]\ttrain's rmse: 3.58019\tvalid's rmse: 3.62348\n",
      "[1500]\ttrain's rmse: 3.52165\tvalid's rmse: 3.58738\n",
      "[2000]\ttrain's rmse: 3.48055\tvalid's rmse: 3.56725\n",
      "[2500]\ttrain's rmse: 3.449\tvalid's rmse: 3.55566\n",
      "[3000]\ttrain's rmse: 3.42031\tvalid's rmse: 3.5463\n",
      "[3500]\ttrain's rmse: 3.39548\tvalid's rmse: 3.53904\n",
      "[4000]\ttrain's rmse: 3.37351\tvalid's rmse: 3.53159\n",
      "[4500]\ttrain's rmse: 3.35393\tvalid's rmse: 3.52702\n",
      "[5000]\ttrain's rmse: 3.33695\tvalid's rmse: 3.52265\n",
      "[5500]\ttrain's rmse: 3.32203\tvalid's rmse: 3.51956\n",
      "[6000]\ttrain's rmse: 3.30605\tvalid's rmse: 3.51775\n",
      "[6500]\ttrain's rmse: 3.2908\tvalid's rmse: 3.51558\n",
      "[7000]\ttrain's rmse: 3.27691\tvalid's rmse: 3.51353\n",
      "[7500]\ttrain's rmse: 3.26337\tvalid's rmse: 3.51151\n",
      "[8000]\ttrain's rmse: 3.25099\tvalid's rmse: 3.5104\n",
      "[8500]\ttrain's rmse: 3.23879\tvalid's rmse: 3.50962\n",
      "[9000]\ttrain's rmse: 3.22704\tvalid's rmse: 3.50892\n",
      "Early stopping, best iteration is:\n",
      "[9365]\ttrain's rmse: 3.21875\tvalid's rmse: 3.50847\n",
      "Fold  5 rmse : 3.508468\n",
      "Full rmse 3.538455\n"
     ]
    }
   ],
   "source": [
    "folds = KFold(n_splits=5, shuffle=True, random_state=1001)\n",
    "# Create arrays and dataframes to store results\n",
    "oof_preds = np.zeros(train.shape[0])\n",
    "sub_preds = np.zeros(test.shape[0])\n",
    "feature_importance_df = pd.DataFrame()\n",
    "    \n",
    "for n_fold, (train_idx, valid_idx) in enumerate(folds.split(train[feats], train['fare_amount'])):\n",
    "    dtrain = lgb.Dataset(data=train[feats].iloc[train_idx], \n",
    "                         label=train['fare_amount'].iloc[train_idx],\n",
    "                         free_raw_data=False)\n",
    "    dvalid = lgb.Dataset(data=train[feats].iloc[valid_idx],\n",
    "                         label=train['fare_amount'].iloc[valid_idx],\n",
    "                         free_raw_data=False)\n",
    "    params = {'boosting_type': 'gbdt',\n",
    "              'objective': 'regression',\n",
    "              'metric':'rmse',\n",
    "              'learning_rate': 0.03,\n",
    "              'num_leaves': 30, \n",
    "              'max_depth': 7,  \n",
    "              'min_child_samples': 70,  \n",
    "              'max_bin': 300,  \n",
    "              'subsample': 1.0,  \n",
    "              'subsample_freq': 1,  \n",
    "              'colsample_bytree': 0.9,  \n",
    "              'min_split_gain': 0.5,\n",
    "              'min_child_weight': 4,\n",
    "              'reg_lambda':0.1,\n",
    "              'reg_alpha': 0.1,\n",
    "              'nthread': 8,\n",
    "              'verbose': -1,}\n",
    "    \n",
    "    clf = lgb.train(params, \n",
    "                    dtrain, \n",
    "                    valid_sets=[dtrain, dvalid], \n",
    "                    valid_names=['train','valid'],\n",
    "                    num_boost_round=10000,\n",
    "                    early_stopping_rounds=125,\n",
    "                    verbose_eval=500)\n",
    "\n",
    "    oof_preds[valid_idx] = clf.predict(train[feats].iloc[valid_idx])\n",
    "    sub_preds += clf.predict(test[feats]) / folds.n_splits\n",
    "\n",
    "    fold_importance_df = pd.DataFrame()\n",
    "    fold_importance_df[\"feature\"] = feats\n",
    "    fold_importance_df[\"importance\"] = clf.feature_importance(importance_type='gain')\n",
    "    fold_importance_df[\"fold\"] = n_fold + 1\n",
    "    feature_importance_df = pd.concat([feature_importance_df, fold_importance_df], axis=0)\n",
    "    print('Fold %2d rmse : %.6f' % (n_fold + 1,mean_squared_error(train['fare_amount'].iloc[valid_idx], oof_preds[valid_idx]) ** .5)) \n",
    "    del clf, dtrain, dvalid\n",
    "    gc.collect()\n",
    "\n",
    "print('Full rmse %.6f' % mean_squared_error(train['fare_amount'], oof_preds)**.5)\n",
    "# Write submission file and plot feature importance\n",
    "sub_df = test[['key']].copy()\n",
    "sub_df['fare_amount'] = sub_preds\n",
    "sub_df[['key', 'fare_amount']].to_csv('submission_lgb2.csv', index= False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAlAAAAJaCAYAAADgeXkjAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\ntS0HAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x5162d4dd0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "cols = feature_importance_df[[\"feature\", \"importance\"]].groupby(\"feature\").mean().sort_values(by=\"importance\", ascending=False).index\n",
    "best_features = feature_importance_df.loc[feature_importance_df.feature.isin(cols)]\n",
    "plt.figure(figsize=(8, 10))\n",
    "sns.barplot(x=\"importance\", y=\"feature\", data=best_features.sort_values(by=\"importance\", ascending=False))\n",
    "plt.title('LightGBM Features (avg over folds)')\n",
    "plt.tight_layout\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# bayesian optimization"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "from bayes_opt import BayesianOptimization"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "feats = [f for f in train.columns if f not in ['key','pickup_datetime','fare_amount']]\n",
    "X=train[feats]\n",
    "y=train['fare_amount']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/shuozhang/anaconda/lib/python2.7/site-packages/sklearn/utils/deprecation.py:52: DeprecationWarning: Class GaussianProcess is deprecated; GaussianProcess was deprecated in version 0.18 and will be removed in 0.20. Use the GaussianProcessRegressor instead.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\u001b[31mInitialization\u001b[0m\n",
      "\u001b[94m----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------\u001b[0m\n",
      " Step |   Time |      Value |   colsample_bytree |   max_bin |   max_depth |   min_child_sample |   min_child_weight |   min_split_gain |   num_leaves |   reg_alpha |   reg_lambda |   subsample |   subsample_freq | \n",
      "[500]\tcv_agg's rmse: 3.7389 + 0.077189\n",
      "[1000]\tcv_agg's rmse: 3.6603 + 0.077709\n",
      "[1500]\tcv_agg's rmse: 3.62421 + 0.0780063\n",
      "[2000]\tcv_agg's rmse: 3.60173 + 0.0798864\n",
      "[2500]\tcv_agg's rmse: 3.58684 + 0.0790663\n",
      "[3000]\tcv_agg's rmse: 3.57732 + 0.0794719\n",
      "[3500]\tcv_agg's rmse: 3.57032 + 0.0797964\n",
      "[4000]\tcv_agg's rmse: 3.56562 + 0.0803206\n",
      "[4500]\tcv_agg's rmse: 3.56139 + 0.0803653\n",
      "[5000]\tcv_agg's rmse: 3.558 + 0.0807158\n",
      "[5500]\tcv_agg's rmse: 3.5551 + 0.0804389\n",
      "[6000]\tcv_agg's rmse: 3.55306 + 0.0803268\n",
      "[6500]\tcv_agg's rmse: 3.55124 + 0.0803134\n",
      "[7000]\tcv_agg's rmse: 3.54947 + 0.0805966\n",
      "[7500]\tcv_agg's rmse: 3.54811 + 0.0806124\n",
      "[8000]\tcv_agg's rmse: 3.54688 + 0.0810139\n",
      "[8500]\tcv_agg's rmse: 3.54555 + 0.0809847\n",
      "    1 | 239m44s | \u001b[35m   9.48186\u001b[0m | \u001b[32m            0.6416\u001b[0m | \u001b[32m 332.7679\u001b[0m | \u001b[32m     7.0947\u001b[0m | \u001b[32m           36.2003\u001b[0m | \u001b[32m            4.0567\u001b[0m | \u001b[32m          0.1421\u001b[0m | \u001b[32m     25.0384\u001b[0m | \u001b[32m     0.6913\u001b[0m | \u001b[32m      0.3254\u001b[0m | \u001b[32m     0.9525\u001b[0m | \u001b[32m          2.6974\u001b[0m | \n",
      "[500]\tcv_agg's rmse: 3.66499 + 0.0781868\n",
      "[1000]\tcv_agg's rmse: 3.6013 + 0.0789194\n",
      "[1500]\tcv_agg's rmse: 3.57554 + 0.0799082\n",
      "[2000]\tcv_agg's rmse: 3.56152 + 0.0802247\n",
      "[2500]\tcv_agg's rmse: 3.55282 + 0.0803058\n",
      "[3000]\tcv_agg's rmse: 3.54712 + 0.0809306\n",
      "[3500]\tcv_agg's rmse: 3.54283 + 0.0809975\n",
      "[4000]\tcv_agg's rmse: 3.54016 + 0.0810851\n",
      "[4500]\tcv_agg's rmse: 3.53821 + 0.0812989\n",
      "[5000]\tcv_agg's rmse: 3.53667 + 0.0813646\n",
      "[5500]\tcv_agg's rmse: 3.53568 + 0.081472\n",
      "[6000]\tcv_agg's rmse: 3.53479 + 0.0818058\n",
      "[6500]\tcv_agg's rmse: 3.53397 + 0.0822143\n",
      "[7000]\tcv_agg's rmse: 3.53367 + 0.0819851\n",
      "    2 | 221m55s |    9.47207 |             0.7358 |  495.1206 |      7.1701 |            90.5120 |            16.4563 |           0.4030 |      51.4093 |      0.6375 |       0.9101 |      0.9870 |           8.7798 | \n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/shuozhang/anaconda/lib/python2.7/site-packages/sklearn/utils/deprecation.py:70: DeprecationWarning: Function l1_cross_distances is deprecated; l1_cross_distances was deprecated in version 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\u001b[31mBayesian Optimization\u001b[0m\n",
      "\u001b[94m----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------\u001b[0m\n",
      " Step |   Time |      Value |   colsample_bytree |   max_bin |   max_depth |   min_child_sample |   min_child_weight |   min_split_gain |   num_leaves |   reg_alpha |   reg_lambda |   subsample |   subsample_freq | \n",
      "[500]\tcv_agg's rmse: 3.70086 + 0.0766031\n",
      "[1000]\tcv_agg's rmse: 3.63561 + 0.0784866\n",
      "[1500]\tcv_agg's rmse: 3.60955 + 0.0781933\n",
      "[2000]\tcv_agg's rmse: 3.59603 + 0.0784326\n",
      "[2500]\tcv_agg's rmse: 3.58726 + 0.0777414\n",
      "[3000]\tcv_agg's rmse: 3.58339 + 0.0781642\n",
      "[3500]\tcv_agg's rmse: 3.58002 + 0.0783934\n",
      "[4000]\tcv_agg's rmse: 3.57766 + 0.0786655\n",
      "[4500]\tcv_agg's rmse: 3.57663 + 0.0781121\n",
      "[5000]\tcv_agg's rmse: 3.57576 + 0.0782674\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/shuozhang/anaconda/lib/python2.7/site-packages/sklearn/utils/deprecation.py:70: DeprecationWarning: Function l1_cross_distances is deprecated; l1_cross_distances was deprecated in version 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "    3 | 114m14s |    9.47593 |             0.7682 |  282.6712 |      6.5581 |            71.7125 |            13.5135 |           0.3629 |      34.8234 |      0.5900 |       0.5564 |      0.6666 |           2.1730 | \n",
      "[500]\tcv_agg's rmse: 3.64535 + 0.076292\n",
      "[1000]\tcv_agg's rmse: 3.59984 + 0.078433\n",
      "[1500]\tcv_agg's rmse: 3.5842 + 0.0784307\n",
      "[2000]\tcv_agg's rmse: 3.5773 + 0.0775631\n",
      "[2500]\tcv_agg's rmse: 3.57363 + 0.077713\n",
      "[3000]\tcv_agg's rmse: 3.57165 + 0.0775132\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/shuozhang/anaconda/lib/python2.7/site-packages/sklearn/utils/deprecation.py:70: DeprecationWarning: Function l1_cross_distances is deprecated; l1_cross_distances was deprecated in version 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "    4 | 74m08s |    9.46552 |             0.9974 |  302.4669 |     -0.6260 |            20.0605 |            16.3854 |           0.2825 |      57.3723 |      0.2619 |       0.0346 |      0.6781 |           6.8627 | \n",
      "[500]\tcv_agg's rmse: 3.61551 + 0.080196\n",
      "[1000]\tcv_agg's rmse: 3.58107 + 0.0798847\n",
      "[1500]\tcv_agg's rmse: 3.57029 + 0.0796853\n",
      "[2000]\tcv_agg's rmse: 3.56565 + 0.0795233\n",
      "[2500]\tcv_agg's rmse: 3.56266 + 0.07895\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/shuozhang/anaconda/lib/python2.7/site-packages/sklearn/utils/deprecation.py:70: DeprecationWarning: Function l1_cross_distances is deprecated; l1_cross_distances was deprecated in version 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "    5 | 72m16s |    9.46267 |             0.6534 |  289.7591 |      0.0118 |            82.2728 |             5.6131 |           0.2826 |      87.8061 |      0.5311 |       0.6326 |      0.6718 |           5.6886 | \n",
      "[500]\tcv_agg's rmse: 4.11906 + 0.0711536\n",
      "[1000]\tcv_agg's rmse: 4.03231 + 0.0728352\n",
      "[1500]\tcv_agg's rmse: 3.98635 + 0.0739083\n",
      "[2000]\tcv_agg's rmse: 3.95512 + 0.0746794\n",
      "[2500]\tcv_agg's rmse: 3.93223 + 0.075267\n",
      "[3000]\tcv_agg's rmse: 3.91398 + 0.0751384\n",
      "[3500]\tcv_agg's rmse: 3.89838 + 0.0749701\n",
      "[4000]\tcv_agg's rmse: 3.88486 + 0.0747855\n",
      "[4500]\tcv_agg's rmse: 3.87396 + 0.0754202\n",
      "[5000]\tcv_agg's rmse: 3.86428 + 0.0754576\n",
      "[5500]\tcv_agg's rmse: 3.85611 + 0.0755697\n",
      "[6000]\tcv_agg's rmse: 3.84874 + 0.0758069\n",
      "[6500]\tcv_agg's rmse: 3.84263 + 0.0758138\n",
      "[7000]\tcv_agg's rmse: 3.83711 + 0.0760047\n",
      "[7500]\tcv_agg's rmse: 3.83202 + 0.0761537\n",
      "[8000]\tcv_agg's rmse: 3.82725 + 0.0761413\n",
      "[8500]\tcv_agg's rmse: 3.82318 + 0.0760159\n",
      "[9000]\tcv_agg's rmse: 3.81952 + 0.0761586\n",
      "[9500]\tcv_agg's rmse: 3.81623 + 0.0762039\n",
      "[10000]\tcv_agg's rmse: 3.81299 + 0.076187\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/shuozhang/anaconda/lib/python2.7/site-packages/sklearn/utils/deprecation.py:70: DeprecationWarning: Function l1_cross_distances is deprecated; l1_cross_distances was deprecated in version 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "    6 | 445m39s | \u001b[35m   9.54276\u001b[0m | \u001b[32m            0.9535\u001b[0m | \u001b[32m 257.4187\u001b[0m | \u001b[32m     1.6330\u001b[0m | \u001b[32m           84.5715\u001b[0m | \u001b[32m            3.9816\u001b[0m | \u001b[32m          0.3573\u001b[0m | \u001b[32m     74.2168\u001b[0m | \u001b[32m     0.2484\u001b[0m | \u001b[32m      0.7279\u001b[0m | \u001b[32m     0.7844\u001b[0m | \u001b[32m          5.1936\u001b[0m | \n",
      "[500]\tcv_agg's rmse: 3.71564 + 0.0783027\n",
      "[1000]\tcv_agg's rmse: 3.64483 + 0.0784131\n",
      "[1500]\tcv_agg's rmse: 3.61261 + 0.0783215\n",
      "[2000]\tcv_agg's rmse: 3.59465 + 0.079101\n",
      "[2500]\tcv_agg's rmse: 3.58395 + 0.07913\n",
      "[3000]\tcv_agg's rmse: 3.57661 + 0.0786683\n",
      "[3500]\tcv_agg's rmse: 3.57183 + 0.0787636\n",
      "[4000]\tcv_agg's rmse: 3.56856 + 0.0792882\n",
      "[4500]\tcv_agg's rmse: 3.56578 + 0.0793105\n",
      "[5000]\tcv_agg's rmse: 3.56374 + 0.0793468\n",
      "[5500]\tcv_agg's rmse: 3.56231 + 0.0793637\n",
      "[6000]\tcv_agg's rmse: 3.56126 + 0.0796651\n",
      "[6500]\tcv_agg's rmse: 3.56025 + 0.0798529\n",
      "[7000]\tcv_agg's rmse: 3.55956 + 0.0799768\n",
      "[7500]\tcv_agg's rmse: 3.55883 + 0.0800363\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/shuozhang/anaconda/lib/python2.7/site-packages/sklearn/utils/deprecation.py:70: DeprecationWarning: Function l1_cross_distances is deprecated; l1_cross_distances was deprecated in version 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "    7 | 174m17s |    9.47672 |             0.9792 |  310.4915 |      7.8375 |            72.5143 |             4.6034 |           0.3077 |      28.4799 |      0.7121 |       0.4914 |      0.8541 |           7.1275 | \n",
      "[500]\tcv_agg's rmse: 4.12866 + 0.0737205\n",
      "[1000]\tcv_agg's rmse: 4.04125 + 0.072624\n",
      "[1500]\tcv_agg's rmse: 3.99459 + 0.0726977\n",
      "[2000]\tcv_agg's rmse: 3.96211 + 0.0722755\n",
      "[2500]\tcv_agg's rmse: 3.938 + 0.0726952\n",
      "[3000]\tcv_agg's rmse: 3.91873 + 0.0732676\n",
      "[3500]\tcv_agg's rmse: 3.90291 + 0.0736872\n",
      "[4000]\tcv_agg's rmse: 3.88958 + 0.0739436\n",
      "[4500]\tcv_agg's rmse: 3.87841 + 0.0742598\n",
      "[5000]\tcv_agg's rmse: 3.86856 + 0.0745951\n",
      "[5500]\tcv_agg's rmse: 3.86031 + 0.0747386\n",
      "[6000]\tcv_agg's rmse: 3.85314 + 0.0749135\n",
      "[6500]\tcv_agg's rmse: 3.84688 + 0.0749186\n",
      "[7000]\tcv_agg's rmse: 3.8415 + 0.0751443\n",
      "[7500]\tcv_agg's rmse: 3.83657 + 0.0754325\n",
      "[8000]\tcv_agg's rmse: 3.83229 + 0.0753873\n",
      "[8500]\tcv_agg's rmse: 3.82843 + 0.0755746\n",
      "[9000]\tcv_agg's rmse: 3.82468 + 0.075794\n",
      "[9500]\tcv_agg's rmse: 3.82148 + 0.076247\n",
      "[10000]\tcv_agg's rmse: 3.8183 + 0.0763523\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/shuozhang/anaconda/lib/python2.7/site-packages/sklearn/utils/deprecation.py:70: DeprecationWarning: Function l1_cross_distances is deprecated; l1_cross_distances was deprecated in version 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "    8 | 190m51s | \u001b[35m   9.54277\u001b[0m | \u001b[32m            0.6385\u001b[0m | \u001b[32m 197.9333\u001b[0m | \u001b[32m     2.0102\u001b[0m | \u001b[32m           25.7341\u001b[0m | \u001b[32m           10.9237\u001b[0m | \u001b[32m          0.7046\u001b[0m | \u001b[32m     63.8524\u001b[0m | \u001b[32m     0.8471\u001b[0m | \u001b[32m      0.9800\u001b[0m | \u001b[32m     0.7159\u001b[0m | \u001b[32m          9.6980\u001b[0m | \n",
      "[500]\tcv_agg's rmse: 3.94372 + 0.0745681\n",
      "[1000]\tcv_agg's rmse: 3.85085 + 0.0741413\n",
      "[1500]\tcv_agg's rmse: 3.80763 + 0.0749693\n",
      "[2000]\tcv_agg's rmse: 3.77958 + 0.075038\n",
      "[2500]\tcv_agg's rmse: 3.75916 + 0.0751144\n",
      "[3000]\tcv_agg's rmse: 3.74388 + 0.0758279\n",
      "[3500]\tcv_agg's rmse: 3.73176 + 0.076083\n",
      "[4000]\tcv_agg's rmse: 3.72183 + 0.0764429\n",
      "[4500]\tcv_agg's rmse: 3.7141 + 0.0762922\n",
      "[5000]\tcv_agg's rmse: 3.70779 + 0.0763317\n",
      "[5500]\tcv_agg's rmse: 3.70263 + 0.076224\n",
      "[6000]\tcv_agg's rmse: 3.69813 + 0.0765907\n",
      "[6500]\tcv_agg's rmse: 3.69411 + 0.0767339\n",
      "[7000]\tcv_agg's rmse: 3.69048 + 0.0767103\n",
      "[7500]\tcv_agg's rmse: 3.68746 + 0.0765894\n",
      "[8000]\tcv_agg's rmse: 3.68489 + 0.0766262\n",
      "[8500]\tcv_agg's rmse: 3.68247 + 0.0768367\n",
      "[9000]\tcv_agg's rmse: 3.68021 + 0.076788\n",
      "[9500]\tcv_agg's rmse: 3.67821 + 0.076913\n",
      "[10000]\tcv_agg's rmse: 3.67642 + 0.0769322\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/shuozhang/anaconda/lib/python2.7/site-packages/sklearn/utils/deprecation.py:70: DeprecationWarning: Function l1_cross_distances is deprecated; l1_cross_distances was deprecated in version 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "    9 | 225m43s |    9.50967 |             0.8049 |  334.5730 |      2.8740 |            39.7750 |            14.7969 |           0.2904 |      33.2033 |      0.5609 |       0.9679 |      0.7745 |           2.8845 | \n",
      "[500]\tcv_agg's rmse: 3.70468 + 0.0781848\n",
      "[1000]\tcv_agg's rmse: 3.63384 + 0.0779759\n",
      "[1500]\tcv_agg's rmse: 3.60173 + 0.0795722\n",
      "[2000]\tcv_agg's rmse: 3.58394 + 0.0797885\n",
      "[2500]\tcv_agg's rmse: 3.57368 + 0.0801375\n",
      "[3000]\tcv_agg's rmse: 3.56667 + 0.0799052\n",
      "[3500]\tcv_agg's rmse: 3.56166 + 0.0803563\n",
      "[4000]\tcv_agg's rmse: 3.55811 + 0.0798937\n",
      "[4500]\tcv_agg's rmse: 3.55526 + 0.079441\n",
      "[5000]\tcv_agg's rmse: 3.5536 + 0.078938\n",
      "[5500]\tcv_agg's rmse: 3.55177 + 0.0789698\n",
      "[6000]\tcv_agg's rmse: 3.55033 + 0.0789378\n",
      "[6500]\tcv_agg's rmse: 3.54912 + 0.0790856\n",
      "[7000]\tcv_agg's rmse: 3.54786 + 0.0794053\n",
      "[7500]\tcv_agg's rmse: 3.54678 + 0.0794196\n",
      "[8000]\tcv_agg's rmse: 3.54651 + 0.079485\n",
      "[8500]\tcv_agg's rmse: 3.54602 + 0.0794933\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/shuozhang/anaconda/lib/python2.7/site-packages/sklearn/utils/deprecation.py:70: DeprecationWarning: Function l1_cross_distances is deprecated; l1_cross_distances was deprecated in version 0.18 and will be removed in 0.20.\n",
      "  warnings.warn(msg, category=DeprecationWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "   10 | 175m47s |    9.47516 |             0.9249 |  396.0655 |     -0.8455 |            23.1236 |            17.3630 |           0.7214 |      28.7484 |      0.9768 |       0.6913 |      0.8895 |           6.4330 | \n"
     ]
    },
    {
     "ename": "AttributeError",
     "evalue": "'BayesianOptimization' object has no attribute 'points_to_csv'",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[0;31mAttributeError\u001b[0m                            Traceback (most recent call last)",
      "\u001b[0;32m<ipython-input-18-6e36b39f0486>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[1;32m     40\u001b[0m     \u001b[0;32mreturn\u001b[0m \u001b[0mlgbBO\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mres\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;34m'max'\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;34m'max_params'\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     41\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 42\u001b[0;31m \u001b[0mopt_params\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mbayes_parameter_opt_lgb\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mX\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0my\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0minit_round\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;36m2\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mopt_round\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;36m8\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mn_folds\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;36m5\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mrandom_seed\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;36m6\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mn_estimators\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;36m10000\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mlearning_rate\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;36m0.03\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0moutput_process\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mTrue\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m",
      "\u001b[0;32m<ipython-input-18-6e36b39f0486>\u001b[0m in \u001b[0;36mbayes_parameter_opt_lgb\u001b[0;34m(X, y, init_round, opt_round, n_folds, random_seed, n_estimators, learning_rate, output_process)\u001b[0m\n\u001b[1;32m     35\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     36\u001b[0m     \u001b[0;31m# output optimization process\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 37\u001b[0;31m     \u001b[0;32mif\u001b[0m \u001b[0moutput_process\u001b[0m\u001b[0;34m==\u001b[0m\u001b[0mTrue\u001b[0m\u001b[0;34m:\u001b[0m \u001b[0mlgbBO\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mpoints_to_csv\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m\"bayes_opt_result.csv\"\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     38\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     39\u001b[0m     \u001b[0;31m# return best parameters\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;31mAttributeError\u001b[0m: 'BayesianOptimization' object has no attribute 'points_to_csv'"
     ]
    }
   ],
   "source": [
    "def bayes_parameter_opt_lgb(X, y, init_round=2, opt_round=8, n_folds=5, random_seed=6, n_estimators=10000, learning_rate=0.03, output_process=True):\n",
    "    # prepare data\n",
    "    train_data = lgb.Dataset(data=X, label=y,free_raw_data=False)\n",
    "    # parameters\n",
    "    def lgb_eval(num_leaves, colsample_bytree, subsample, max_depth, reg_lambda, reg_alpha, min_split_gain, min_child_weight, \n",
    "                min_child_sample, max_bin, subsample_freq):\n",
    "        params = {'objective':'regression','boosting_type': 'gbdt','nthread': 4, 'verbose': -1,\\\n",
    "                  'num_boost_round': n_estimators, 'learning_rate':learning_rate, \\\n",
    "                  'early_stopping_round':125}\n",
    "        params['subsample_freq']=int(round(subsample_freq))\n",
    "        params['min_child_sample']=int(round(min_child_sample))\n",
    "        params['max_bin']=int(round(max_bin))\n",
    "        params[\"num_leaves\"] = int(round(num_leaves))\n",
    "        params['colsample_bytree'] = max(min(colsample_bytree, 1), 0)\n",
    "        params['subsample'] = max(min(subsample, 1), 0)\n",
    "        params['max_depth'] = int(round(max_depth))\n",
    "        params['reg_lambda'] = max(reg_lambda, 0)\n",
    "        params['reg_alpha'] = max(reg_alpha, 0)\n",
    "        params['min_split_gain'] = min_split_gain\n",
    "        params['min_child_weight'] = min_child_weight\n",
    "        cv_result = lgb.cv(params, train_data, nfold=n_folds, seed=random_seed, stratified=True, verbose_eval=500, metrics=['rmse'])\n",
    "        return max(cv_result['rmse-mean'])\n",
    "    # range \n",
    "    lgbBO = BayesianOptimization(lgb_eval, {'num_leaves': (20, 100),\n",
    "                                            'colsample_bytree': (0.6, 1.0),\n",
    "                                            'subsample': (0.6, 1.0),\n",
    "                                            'max_depth': (-1, 8),\n",
    "                                            'reg_lambda': (0, 1),\n",
    "                                            'reg_alpha': (0, 1),\n",
    "                                            'min_child_sample':(20,100),\n",
    "                                            'max_bin':(180,500),\n",
    "                                            'subsample_freq':(1,10),\n",
    "                                            'min_split_gain': (0.1, 0.8),\n",
    "                                            'min_child_weight': (3, 20)})\n",
    "    # optimize\n",
    "    lgbBO.maximize(init_points=init_round, n_iter=opt_round)\n",
    "    \n",
    "    #return lgbBO.res['max']['max_params']\n",
    "\n",
    "opt_params = bayes_parameter_opt_lgb(X, y, init_round=2, opt_round=8, n_folds=5, random_seed=6, n_estimators=10000, learning_rate=0.03,output_process=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training until validation scores don't improve for 125 rounds.\n",
      "[500]\ttrain's rmse: 3.70469\tvalid's rmse: 3.72333\n",
      "[1000]\ttrain's rmse: 3.59408\tvalid's rmse: 3.64641\n",
      "[1500]\ttrain's rmse: 3.53108\tvalid's rmse: 3.61018\n",
      "[2000]\ttrain's rmse: 3.48629\tvalid's rmse: 3.58888\n",
      "[2500]\ttrain's rmse: 3.45111\tvalid's rmse: 3.57509\n",
      "[3000]\ttrain's rmse: 3.42156\tvalid's rmse: 3.56619\n",
      "[3500]\ttrain's rmse: 3.39559\tvalid's rmse: 3.55988\n",
      "[4000]\ttrain's rmse: 3.37225\tvalid's rmse: 3.55506\n",
      "[4500]\ttrain's rmse: 3.3516\tvalid's rmse: 3.55139\n",
      "[5000]\ttrain's rmse: 3.33264\tvalid's rmse: 3.54828\n",
      "[5500]\ttrain's rmse: 3.31405\tvalid's rmse: 3.54517\n",
      "[6000]\ttrain's rmse: 3.29739\tvalid's rmse: 3.54332\n",
      "[6500]\ttrain's rmse: 3.28121\tvalid's rmse: 3.54136\n",
      "Early stopping, best iteration is:\n",
      "[6848]\ttrain's rmse: 3.27041\tvalid's rmse: 3.54051\n",
      "Fold  1 rmse : 3.540509\n",
      "Training until validation scores don't improve for 125 rounds.\n",
      "[500]\ttrain's rmse: 3.67881\tvalid's rmse: 3.85635\n",
      "[1000]\ttrain's rmse: 3.57265\tvalid's rmse: 3.77658\n",
      "[1500]\ttrain's rmse: 3.51325\tvalid's rmse: 3.74298\n",
      "[2000]\ttrain's rmse: 3.46928\tvalid's rmse: 3.72161\n",
      "[2500]\ttrain's rmse: 3.4335\tvalid's rmse: 3.70677\n",
      "[3000]\ttrain's rmse: 3.40558\tvalid's rmse: 3.6972\n",
      "[3500]\ttrain's rmse: 3.38116\tvalid's rmse: 3.69144\n",
      "[4000]\ttrain's rmse: 3.35936\tvalid's rmse: 3.68646\n",
      "[4500]\ttrain's rmse: 3.34004\tvalid's rmse: 3.68303\n",
      "[5000]\ttrain's rmse: 3.32151\tvalid's rmse: 3.67997\n",
      "[5500]\ttrain's rmse: 3.30371\tvalid's rmse: 3.67695\n",
      "[6000]\ttrain's rmse: 3.28874\tvalid's rmse: 3.67459\n",
      "[6500]\ttrain's rmse: 3.27394\tvalid's rmse: 3.67268\n",
      "[7000]\ttrain's rmse: 3.2598\tvalid's rmse: 3.67156\n",
      "[7500]\ttrain's rmse: 3.24688\tvalid's rmse: 3.67031\n",
      "[8000]\ttrain's rmse: 3.2337\tvalid's rmse: 3.66894\n",
      "[8500]\ttrain's rmse: 3.22122\tvalid's rmse: 3.66757\n",
      "[9000]\ttrain's rmse: 3.20899\tvalid's rmse: 3.66673\n",
      "[9500]\ttrain's rmse: 3.19788\tvalid's rmse: 3.66609\n",
      "Early stopping, best iteration is:\n",
      "[9493]\ttrain's rmse: 3.19803\tvalid's rmse: 3.66604\n",
      "Fold  2 rmse : 3.666044\n",
      "Training until validation scores don't improve for 125 rounds.\n",
      "[500]\ttrain's rmse: 3.71577\tvalid's rmse: 3.68813\n",
      "[1000]\ttrain's rmse: 3.60294\tvalid's rmse: 3.61029\n",
      "[1500]\ttrain's rmse: 3.5384\tvalid's rmse: 3.5744\n",
      "[2000]\ttrain's rmse: 3.49276\tvalid's rmse: 3.55449\n",
      "[2500]\ttrain's rmse: 3.45574\tvalid's rmse: 3.5408\n",
      "[3000]\ttrain's rmse: 3.42503\tvalid's rmse: 3.53226\n",
      "[3500]\ttrain's rmse: 3.39851\tvalid's rmse: 3.52718\n",
      "[4000]\ttrain's rmse: 3.37501\tvalid's rmse: 3.52223\n",
      "[4500]\ttrain's rmse: 3.35437\tvalid's rmse: 3.51945\n",
      "[5000]\ttrain's rmse: 3.3355\tvalid's rmse: 3.51668\n",
      "[5500]\ttrain's rmse: 3.31751\tvalid's rmse: 3.51448\n",
      "[6000]\ttrain's rmse: 3.30063\tvalid's rmse: 3.51305\n",
      "[6500]\ttrain's rmse: 3.28402\tvalid's rmse: 3.51144\n",
      "[7000]\ttrain's rmse: 3.26935\tvalid's rmse: 3.51029\n",
      "Early stopping, best iteration is:\n",
      "[7219]\ttrain's rmse: 3.26309\tvalid's rmse: 3.50977\n",
      "Fold  3 rmse : 3.509773\n",
      "Training until validation scores don't improve for 125 rounds.\n",
      "[500]\ttrain's rmse: 3.71248\tvalid's rmse: 3.71824\n",
      "[1000]\ttrain's rmse: 3.60132\tvalid's rmse: 3.63185\n",
      "[1500]\ttrain's rmse: 3.53917\tvalid's rmse: 3.59284\n",
      "[2000]\ttrain's rmse: 3.49247\tvalid's rmse: 3.56976\n",
      "[2500]\ttrain's rmse: 3.45615\tvalid's rmse: 3.55556\n",
      "[3000]\ttrain's rmse: 3.42557\tvalid's rmse: 3.54571\n",
      "[3500]\ttrain's rmse: 3.3991\tvalid's rmse: 3.53907\n",
      "[4000]\ttrain's rmse: 3.37551\tvalid's rmse: 3.53299\n",
      "[4500]\ttrain's rmse: 3.35445\tvalid's rmse: 3.52886\n",
      "[5000]\ttrain's rmse: 3.33471\tvalid's rmse: 3.52487\n",
      "[5500]\ttrain's rmse: 3.31767\tvalid's rmse: 3.52176\n",
      "[6000]\ttrain's rmse: 3.30139\tvalid's rmse: 3.51955\n",
      "[6500]\ttrain's rmse: 3.28525\tvalid's rmse: 3.51733\n",
      "[7000]\ttrain's rmse: 3.27042\tvalid's rmse: 3.51502\n",
      "[7500]\ttrain's rmse: 3.25622\tvalid's rmse: 3.51346\n",
      "[8000]\ttrain's rmse: 3.24324\tvalid's rmse: 3.51212\n",
      "[8500]\ttrain's rmse: 3.23023\tvalid's rmse: 3.5111\n",
      "[9000]\ttrain's rmse: 3.21788\tvalid's rmse: 3.51035\n",
      "Early stopping, best iteration is:\n",
      "[8950]\ttrain's rmse: 3.21914\tvalid's rmse: 3.51016\n",
      "Fold  4 rmse : 3.510164\n",
      "Training until validation scores don't improve for 125 rounds.\n",
      "[500]\ttrain's rmse: 3.71166\tvalid's rmse: 3.70082\n",
      "[1000]\ttrain's rmse: 3.60438\tvalid's rmse: 3.62015\n",
      "[1500]\ttrain's rmse: 3.54484\tvalid's rmse: 3.58172\n",
      "[2000]\ttrain's rmse: 3.49903\tvalid's rmse: 3.55849\n",
      "[2500]\ttrain's rmse: 3.46344\tvalid's rmse: 3.54343\n",
      "[3000]\ttrain's rmse: 3.43386\tvalid's rmse: 3.5335\n",
      "[3500]\ttrain's rmse: 3.40807\tvalid's rmse: 3.52494\n",
      "[4000]\ttrain's rmse: 3.38587\tvalid's rmse: 3.51925\n",
      "[4500]\ttrain's rmse: 3.36517\tvalid's rmse: 3.51359\n",
      "[5000]\ttrain's rmse: 3.34648\tvalid's rmse: 3.50986\n",
      "[5500]\ttrain's rmse: 3.32902\tvalid's rmse: 3.50676\n",
      "[6000]\ttrain's rmse: 3.31151\tvalid's rmse: 3.50436\n",
      "[6500]\ttrain's rmse: 3.29683\tvalid's rmse: 3.50256\n",
      "[7000]\ttrain's rmse: 3.28273\tvalid's rmse: 3.50069\n",
      "[7500]\ttrain's rmse: 3.26868\tvalid's rmse: 3.49836\n",
      "[8000]\ttrain's rmse: 3.25563\tvalid's rmse: 3.4972\n",
      "[8500]\ttrain's rmse: 3.24273\tvalid's rmse: 3.49602\n",
      "[9000]\ttrain's rmse: 3.22987\tvalid's rmse: 3.49492\n",
      "[9500]\ttrain's rmse: 3.21736\tvalid's rmse: 3.49337\n",
      "[10000]\ttrain's rmse: 3.20644\tvalid's rmse: 3.49209\n",
      "Fold  5 rmse : 3.492090\n",
      "Full rmse 3.544278\n"
     ]
    }
   ],
   "source": [
    "folds = KFold(n_splits=5, shuffle=True, random_state=2001)\n",
    "# Create arrays and dataframes to store results\n",
    "oof_preds = np.zeros(train.shape[0])\n",
    "sub_preds = np.zeros(test.shape[0])\n",
    "feature_importance_df = pd.DataFrame()\n",
    "    \n",
    "for n_fold, (train_idx, valid_idx) in enumerate(folds.split(train[feats], train['fare_amount'])):\n",
    "    dtrain = lgb.Dataset(data=train[feats].iloc[train_idx], \n",
    "                         label=train['fare_amount'].iloc[train_idx],\n",
    "                         free_raw_data=False)\n",
    "    dvalid = lgb.Dataset(data=train[feats].iloc[valid_idx],\n",
    "                         label=train['fare_amount'].iloc[valid_idx],\n",
    "                         free_raw_data=False)\n",
    "    params = {'boosting_type': 'gbdt',\n",
    "              'objective': 'regression',\n",
    "              'metric':'rmse',\n",
    "              'learning_rate': 0.03,\n",
    "              'num_leaves': 25, \n",
    "              'max_depth': 7,  \n",
    "              'min_child_samples': 36,  \n",
    "              'max_bin': 333,  \n",
    "              'subsample': 0.9525,  \n",
    "              'subsample_freq': 3,  \n",
    "              'colsample_bytree': 0.6416,  \n",
    "              'min_split_gain': 0.1421,\n",
    "              'min_child_weight': 4.0567,\n",
    "              'reg_lambda':0.3254,\n",
    "              'reg_alpha': 0.6913,\n",
    "              'nthread': 8,\n",
    "              'verbose': -1,}\n",
    "    \n",
    "    clf = lgb.train(params, \n",
    "                    dtrain, \n",
    "                    valid_sets=[dtrain, dvalid], \n",
    "                    valid_names=['train','valid'],\n",
    "                    num_boost_round=10000,\n",
    "                    early_stopping_rounds=125,\n",
    "                    verbose_eval=500)\n",
    "\n",
    "    oof_preds[valid_idx] = clf.predict(train[feats].iloc[valid_idx])\n",
    "    sub_preds += clf.predict(test[feats]) / folds.n_splits\n",
    "\n",
    "    fold_importance_df = pd.DataFrame()\n",
    "    fold_importance_df[\"feature\"] = feats\n",
    "    fold_importance_df[\"importance\"] = clf.feature_importance(importance_type='gain')\n",
    "    fold_importance_df[\"fold\"] = n_fold + 1\n",
    "    feature_importance_df = pd.concat([feature_importance_df, fold_importance_df], axis=0)\n",
    "    print('Fold %2d rmse : %.6f' % (n_fold + 1,mean_squared_error(train['fare_amount'].iloc[valid_idx], oof_preds[valid_idx]) ** .5)) \n",
    "    del clf, dtrain, dvalid\n",
    "    gc.collect()\n",
    "\n",
    "print('Full rmse %.6f' % mean_squared_error(train['fare_amount'], oof_preds)**.5)\n",
    "# Write submission file and plot feature importance\n",
    "sub_df = test[['key']].copy()\n",
    "sub_df['fare_amount'] = sub_preds\n",
    "sub_df[['key', 'fare_amount']].to_csv('submission_lgb_bayesian.csv', index= False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# since distance is very important, we need to add more distance realted features\n",
    "# also some time feature contribute too little, we will consider to remove them from features\n",
    "# anything else: some numeric feature to categorical feature"
   ]
  }
 ],
 "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": "2.7.15"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}