Car_Accident_Notebook.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "3e5e9fcc",
   "metadata": {},
   "source": [
    "# <center><em>IITH Hackathon: Driver Fault Classification</em></center>"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "139e0c30",
   "metadata": {},
   "source": [
    "___\n",
    "\n",
    "<a href='https://www.kaggle.com/c/Foml-2021/overview'><img src='header.png'/></a>\n",
    "___"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "313f3fe3",
   "metadata": {},
   "source": [
    "### Importing the necessary modules"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "983990d7",
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import numpy as np\n",
    "import seaborn as sns\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "import warnings\n",
    "warnings.filterwarnings(\"ignore\")\n",
    "\n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn.ensemble import GradientBoostingClassifier\n",
    "from sklearn.ensemble import RandomForestClassifier \n",
    "from sklearn.metrics import accuracy_score\n",
    "import xgboost as xgb"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "fc21f570",
   "metadata": {},
   "source": [
    "### Loading the training data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "268ddfba",
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Report Number</th>\n",
       "      <th>Local Case Number</th>\n",
       "      <th>Agency Name</th>\n",
       "      <th>ACRS Report Type</th>\n",
       "      <th>Crash Date/Time</th>\n",
       "      <th>Route Type</th>\n",
       "      <th>Road Name</th>\n",
       "      <th>Cross-Street Type</th>\n",
       "      <th>Cross-Street Name</th>\n",
       "      <th>Off-Road Description</th>\n",
       "      <th>...</th>\n",
       "      <th>Driverless Vehicle</th>\n",
       "      <th>Parked Vehicle</th>\n",
       "      <th>Vehicle Year</th>\n",
       "      <th>Vehicle Make</th>\n",
       "      <th>Vehicle Model</th>\n",
       "      <th>Equipment Problems</th>\n",
       "      <th>Latitude</th>\n",
       "      <th>Longitude</th>\n",
       "      <th>Location</th>\n",
       "      <th>Fault</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>MP060D</td>\n",
       "      <td>10196</td>\n",
       "      <td>Montgomery County Police</td>\n",
       "      <td>Injury Crash</td>\n",
       "      <td>2017-06-05 16:27:00</td>\n",
       "      <td>Maryland (State)</td>\n",
       "      <td>OLD HUNDRED RD</td>\n",
       "      <td>County</td>\n",
       "      <td>THURSTON RD</td>\n",
       "      <td>12305 SHOREFIELD RD</td>\n",
       "      <td>...</td>\n",
       "      <td>No</td>\n",
       "      <td>No</td>\n",
       "      <td>2005</td>\n",
       "      <td>GMC</td>\n",
       "      <td>TRUCK</td>\n",
       "      <td>NO MISUSE</td>\n",
       "      <td>39.263378</td>\n",
       "      <td>-77.344203</td>\n",
       "      <td>3.6383 7.4233</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>MP000X</td>\n",
       "      <td>20080</td>\n",
       "      <td>Montgomery County Police</td>\n",
       "      <td>Injury Crash</td>\n",
       "      <td>2020-02-22 10:00:00</td>\n",
       "      <td>County</td>\n",
       "      <td>JANET RD</td>\n",
       "      <td>County</td>\n",
       "      <td>FLACK ST</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>No</td>\n",
       "      <td>Yes</td>\n",
       "      <td>2018</td>\n",
       "      <td>FORD</td>\n",
       "      <td>FIESTA</td>\n",
       "      <td>NO MISUSE</td>\n",
       "      <td>39.068913</td>\n",
       "      <td>-77.063227</td>\n",
       "      <td>3.0862 7.324</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>MP070N</td>\n",
       "      <td>10103</td>\n",
       "      <td>Montgomery County Police</td>\n",
       "      <td>Property Damage Crash</td>\n",
       "      <td>2017-06-11 08:21:00</td>\n",
       "      <td>Maryland (State)</td>\n",
       "      <td>COLESVILLE RD</td>\n",
       "      <td>Maryland (State)</td>\n",
       "      <td>STRUC #15082</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>No</td>\n",
       "      <td>No</td>\n",
       "      <td>2016</td>\n",
       "      <td>NEW FLYER</td>\n",
       "      <td>TBU</td>\n",
       "      <td>NO MISUSE</td>\n",
       "      <td>38.993578</td>\n",
       "      <td>-77.032718</td>\n",
       "      <td>3.9583 7.3783</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>MP130Y</td>\n",
       "      <td>10208</td>\n",
       "      <td>Montgomery County Police</td>\n",
       "      <td>Injury Crash</td>\n",
       "      <td>2019-05-10 07:30:00</td>\n",
       "      <td>County</td>\n",
       "      <td>PARKLAND DR</td>\n",
       "      <td>County</td>\n",
       "      <td>FRANKFORT DR</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>No</td>\n",
       "      <td>No</td>\n",
       "      <td>2001</td>\n",
       "      <td>CHEV</td>\n",
       "      <td>SUBURBAN</td>\n",
       "      <td>OTHER</td>\n",
       "      <td>39.083480</td>\n",
       "      <td>-77.086933</td>\n",
       "      <td>3.84,-70633</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>MP770Y</td>\n",
       "      <td>1091</td>\n",
       "      <td>Montgomery County Police</td>\n",
       "      <td>Injury Crash</td>\n",
       "      <td>2016-08-02 17:00:00</td>\n",
       "      <td>Maryland (State)</td>\n",
       "      <td>OLD GEORGETOWN RD</td>\n",
       "      <td>County</td>\n",
       "      <td>MCKINLEY ST</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>No</td>\n",
       "      <td>No</td>\n",
       "      <td>2013</td>\n",
       "      <td>KIA</td>\n",
       "      <td>RIO</td>\n",
       "      <td>NO MISUSE</td>\n",
       "      <td>38.996942</td>\n",
       "      <td>-77.109168</td>\n",
       "      <td>3.9917 7.0183</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 42 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "  Report Number Local Case Number               Agency Name  \\\n",
       "0        MP060D             10196  Montgomery County Police   \n",
       "1        MP000X             20080  Montgomery County Police   \n",
       "2        MP070N             10103  Montgomery County Police   \n",
       "3        MP130Y             10208  Montgomery County Police   \n",
       "4        MP770Y              1091  Montgomery County Police   \n",
       "\n",
       "        ACRS Report Type     Crash Date/Time        Route Type  \\\n",
       "0           Injury Crash 2017-06-05 16:27:00  Maryland (State)   \n",
       "1           Injury Crash 2020-02-22 10:00:00            County   \n",
       "2  Property Damage Crash 2017-06-11 08:21:00  Maryland (State)   \n",
       "3           Injury Crash 2019-05-10 07:30:00            County   \n",
       "4           Injury Crash 2016-08-02 17:00:00  Maryland (State)   \n",
       "\n",
       "           Road Name Cross-Street Type Cross-Street Name Off-Road Description  \\\n",
       "0     OLD HUNDRED RD            County       THURSTON RD  12305 SHOREFIELD RD   \n",
       "1           JANET RD            County          FLACK ST                  NaN   \n",
       "2      COLESVILLE RD  Maryland (State)      STRUC #15082                  NaN   \n",
       "3        PARKLAND DR            County      FRANKFORT DR                  NaN   \n",
       "4  OLD GEORGETOWN RD            County       MCKINLEY ST                  NaN   \n",
       "\n",
       "   ... Driverless Vehicle Parked Vehicle Vehicle Year Vehicle Make  \\\n",
       "0  ...                 No             No         2005          GMC   \n",
       "1  ...                 No            Yes         2018         FORD   \n",
       "2  ...                 No             No         2016    NEW FLYER   \n",
       "3  ...                 No             No         2001         CHEV   \n",
       "4  ...                 No             No         2013          KIA   \n",
       "\n",
       "  Vehicle Model Equipment Problems   Latitude  Longitude       Location Fault  \n",
       "0         TRUCK          NO MISUSE  39.263378 -77.344203  3.6383 7.4233     1  \n",
       "1        FIESTA          NO MISUSE  39.068913 -77.063227   3.0862 7.324     1  \n",
       "2           TBU          NO MISUSE  38.993578 -77.032718  3.9583 7.3783     0  \n",
       "3      SUBURBAN              OTHER  39.083480 -77.086933    3.84,-70633     1  \n",
       "4           RIO          NO MISUSE  38.996942 -77.109168  3.9917 7.0183     0  \n",
       "\n",
       "[5 rows x 42 columns]"
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df = pd.read_csv(\"train_accident.csv\",parse_dates = ['Crash Date/Time'])\n",
    "df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "b5940245",
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(51490, 42)"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "e27ea783",
   "metadata": {},
   "source": [
    "### Pre-processing and EDA of the training data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "97cfad25",
   "metadata": {
    "scrolled": false
   },
   "outputs": [],
   "source": [
    "def extract(data):\n",
    "    return data[:1]\n",
    "\n",
    "df['Report Category'] = df['Report Number'].apply(extract)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "f5ebc5ec",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "M    45104\n",
       "D     3835\n",
       "E     2208\n",
       "H      343\n",
       "Name: Report Category, dtype: int64"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df['Report Category'].value_counts()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "6892b4c6",
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Report Number                         0\n",
       "Local Case Number                  6453\n",
       "Agency Name                           0\n",
       "ACRS Report Type                      0\n",
       "Crash Date/Time                       0\n",
       "Route Type                         4968\n",
       "Road Name                          4502\n",
       "Cross-Street Type                  5019\n",
       "Cross-Street Name                  4522\n",
       "Off-Road Description              44540\n",
       "Municipality                      43923\n",
       "Related Non-Motorist              47842\n",
       "Collision Type                      299\n",
       "Weather                            4029\n",
       "Surface Condition                  5879\n",
       "Light                               704\n",
       "Traffic Control                    7774\n",
       "Driver Substance Abuse             9095\n",
       "Non-Motorist Substance Abuse      48106\n",
       "Person ID                             0\n",
       "Injury Severity                       0\n",
       "Circumstance                      39738\n",
       "Drivers License State              2528\n",
       "Vehicle ID                            0\n",
       "Vehicle Damage Extent               397\n",
       "Vehicle First Impact Location       222\n",
       "Vehicle Second Impact Location      234\n",
       "Vehicle Body Type                   785\n",
       "Vehicle Movement                    198\n",
       "Vehicle Continuing Dir             1129\n",
       "Vehicle Going Dir                  1113\n",
       "Speed Limit                           0\n",
       "Driverless Vehicle                    0\n",
       "Parked Vehicle                        0\n",
       "Vehicle Year                          0\n",
       "Vehicle Make                          8\n",
       "Vehicle Model                        19\n",
       "Equipment Problems                10666\n",
       "Latitude                              0\n",
       "Longitude                             0\n",
       "Location                              0\n",
       "Fault                                 0\n",
       "Report Category                       0\n",
       "dtype: int64"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.isnull().sum()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "a671af26",
   "metadata": {},
   "outputs": [],
   "source": [
    "# dropped because of high occurence of nan values\n",
    "df.drop(['Off-Road Description','Municipality','Related Non-Motorist','Non-Motorist Substance Abuse','Circumstance','Person ID','Vehicle ID'], axis = 1, inplace = True)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "70604eb8",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(51490, 36)"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.shape   "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "ebe6a65c",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Report Number                         0\n",
       "Local Case Number                  6453\n",
       "Agency Name                           0\n",
       "ACRS Report Type                      0\n",
       "Crash Date/Time                       0\n",
       "Route Type                         4968\n",
       "Road Name                          4502\n",
       "Cross-Street Type                  5019\n",
       "Cross-Street Name                  4522\n",
       "Collision Type                      299\n",
       "Weather                            4029\n",
       "Surface Condition                  5879\n",
       "Light                               704\n",
       "Traffic Control                    7774\n",
       "Driver Substance Abuse             9095\n",
       "Injury Severity                       0\n",
       "Drivers License State              2528\n",
       "Vehicle Damage Extent               397\n",
       "Vehicle First Impact Location       222\n",
       "Vehicle Second Impact Location      234\n",
       "Vehicle Body Type                   785\n",
       "Vehicle Movement                    198\n",
       "Vehicle Continuing Dir             1129\n",
       "Vehicle Going Dir                  1113\n",
       "Speed Limit                           0\n",
       "Driverless Vehicle                    0\n",
       "Parked Vehicle                        0\n",
       "Vehicle Year                          0\n",
       "Vehicle Make                          8\n",
       "Vehicle Model                        19\n",
       "Equipment Problems                10666\n",
       "Latitude                              0\n",
       "Longitude                             0\n",
       "Location                              0\n",
       "Fault                                 0\n",
       "Report Category                       0\n",
       "dtype: int64"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.isnull().sum()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "6c046b3b",
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize = (10,4))\n",
    "sns.countplot(data = df, x = 'Report Category', hue = 'Fault')\n",
    "#plt.ylim(0,500)\n",
    "plt.xticks(rotation = 90)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "4a342464",
   "metadata": {},
   "outputs": [],
   "source": [
    "df['Crash_day'] = df['Crash Date/Time'].dt.day\n",
    "df['Crash_hour'] = df['Crash Date/Time'].dt.hour\n",
    "df['Crash_month'] = df['Crash Date/Time'].dt.month\n",
    "df['Crash_year'] = df['Crash Date/Time'].dt.year\n",
    "df['Crash_day_of_week'] = df['Crash Date/Time'].dt.dayofweek\n",
    "df.drop('Crash Date/Time',axis='columns', inplace=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "c93f32fa",
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "sns.barplot(data = df, x = df['Location'].value_counts()[:10].index, y = df['Location'].value_counts()[:10].values)\n",
    "plt.xticks(rotation = 90)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "f6958eab",
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# almost balanced class label\n",
    "sns.countplot(data = df, x = 'Fault')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "721feffa",
   "metadata": {},
   "outputs": [],
   "source": [
    "# let's check latitude and longitude value ranges"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "id": "c8b47a48",
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Report Number</th>\n",
       "      <th>Local Case Number</th>\n",
       "      <th>Agency Name</th>\n",
       "      <th>ACRS Report Type</th>\n",
       "      <th>Route Type</th>\n",
       "      <th>Road Name</th>\n",
       "      <th>Cross-Street Type</th>\n",
       "      <th>Cross-Street Name</th>\n",
       "      <th>Collision Type</th>\n",
       "      <th>Weather</th>\n",
       "      <th>...</th>\n",
       "      <th>Latitude</th>\n",
       "      <th>Longitude</th>\n",
       "      <th>Location</th>\n",
       "      <th>Fault</th>\n",
       "      <th>Report Category</th>\n",
       "      <th>Crash_day</th>\n",
       "      <th>Crash_hour</th>\n",
       "      <th>Crash_month</th>\n",
       "      <th>Crash_year</th>\n",
       "      <th>Crash_day_of_week</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>0 rows × 40 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "Empty DataFrame\n",
       "Columns: [Report Number, Local Case Number, Agency Name, ACRS Report Type, Route Type, Road Name, Cross-Street Type, Cross-Street Name, Collision Type, Weather, Surface Condition, Light, Traffic Control, Driver Substance Abuse, Injury Severity, Drivers License State, Vehicle Damage Extent, Vehicle First Impact Location, Vehicle Second Impact Location, Vehicle Body Type, Vehicle Movement, Vehicle Continuing Dir, Vehicle Going Dir, Speed Limit, Driverless Vehicle, Parked Vehicle, Vehicle Year, Vehicle Make, Vehicle Model, Equipment Problems, Latitude, Longitude, Location, Fault, Report Category, Crash_day, Crash_hour, Crash_month, Crash_year, Crash_day_of_week]\n",
       "Index: []\n",
       "\n",
       "[0 rows x 40 columns]"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df[(df['Latitude'] < -90) | (df['Latitude'] > 90)] "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "id": "93542a20",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Report Number</th>\n",
       "      <th>Local Case Number</th>\n",
       "      <th>Agency Name</th>\n",
       "      <th>ACRS Report Type</th>\n",
       "      <th>Route Type</th>\n",
       "      <th>Road Name</th>\n",
       "      <th>Cross-Street Type</th>\n",
       "      <th>Cross-Street Name</th>\n",
       "      <th>Collision Type</th>\n",
       "      <th>Weather</th>\n",
       "      <th>...</th>\n",
       "      <th>Latitude</th>\n",
       "      <th>Longitude</th>\n",
       "      <th>Location</th>\n",
       "      <th>Fault</th>\n",
       "      <th>Report Category</th>\n",
       "      <th>Crash_day</th>\n",
       "      <th>Crash_hour</th>\n",
       "      <th>Crash_month</th>\n",
       "      <th>Crash_year</th>\n",
       "      <th>Crash_day_of_week</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>0 rows × 40 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "Empty DataFrame\n",
       "Columns: [Report Number, Local Case Number, Agency Name, ACRS Report Type, Route Type, Road Name, Cross-Street Type, Cross-Street Name, Collision Type, Weather, Surface Condition, Light, Traffic Control, Driver Substance Abuse, Injury Severity, Drivers License State, Vehicle Damage Extent, Vehicle First Impact Location, Vehicle Second Impact Location, Vehicle Body Type, Vehicle Movement, Vehicle Continuing Dir, Vehicle Going Dir, Speed Limit, Driverless Vehicle, Parked Vehicle, Vehicle Year, Vehicle Make, Vehicle Model, Equipment Problems, Latitude, Longitude, Location, Fault, Report Category, Crash_day, Crash_hour, Crash_month, Crash_year, Crash_day_of_week]\n",
       "Index: []\n",
       "\n",
       "[0 rows x 40 columns]"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df[(df['Longitude'] < -180) | (df['Longitude'] > 180)] "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "id": "ca3e738f",
   "metadata": {},
   "outputs": [],
   "source": [
    "df['Location'] = df['Location'].str.replace(',',' ')\n",
    "new = df['Location'].str.split(\" \", n = 1, expand = True)\n",
    "\n",
    "df['Location_x'] = new[0]\n",
    "df['Location_y'] = new[1]\n",
    "df['Location_x'] = df['Location_x'].astype(float)\n",
    "df['Location_y'] = df['Location_y'].astype(float)\n",
    "\n",
    "df.drop('Location', axis = 1, inplace = True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "id": "28f30b17",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(51490, 41)"
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "id": "69f98787",
   "metadata": {},
   "outputs": [],
   "source": [
    "drop_columns = ['Report Number']\n",
    "\n",
    "# drivers license state, local case number, report number removed"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "id": "a5791529",
   "metadata": {},
   "outputs": [],
   "source": [
    "df.drop(df[drop_columns], axis = 1, inplace = True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "id": "8f29a3e9",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(51490, 40)"
      ]
     },
     "execution_count": 21,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "id": "109e2eec",
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Local Case Number</th>\n",
       "      <th>Agency Name</th>\n",
       "      <th>ACRS Report Type</th>\n",
       "      <th>Route Type</th>\n",
       "      <th>Road Name</th>\n",
       "      <th>Cross-Street Type</th>\n",
       "      <th>Cross-Street Name</th>\n",
       "      <th>Collision Type</th>\n",
       "      <th>Weather</th>\n",
       "      <th>Surface Condition</th>\n",
       "      <th>...</th>\n",
       "      <th>Longitude</th>\n",
       "      <th>Fault</th>\n",
       "      <th>Report Category</th>\n",
       "      <th>Crash_day</th>\n",
       "      <th>Crash_hour</th>\n",
       "      <th>Crash_month</th>\n",
       "      <th>Crash_year</th>\n",
       "      <th>Crash_day_of_week</th>\n",
       "      <th>Location_x</th>\n",
       "      <th>Location_y</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>10196</td>\n",
       "      <td>Montgomery County Police</td>\n",
       "      <td>Injury Crash</td>\n",
       "      <td>Maryland (State)</td>\n",
       "      <td>OLD HUNDRED RD</td>\n",
       "      <td>County</td>\n",
       "      <td>THURSTON RD</td>\n",
       "      <td>SINGLE VEHICLE</td>\n",
       "      <td>CLEAR</td>\n",
       "      <td>WET</td>\n",
       "      <td>...</td>\n",
       "      <td>-77.344203</td>\n",
       "      <td>1</td>\n",
       "      <td>M</td>\n",
       "      <td>5</td>\n",
       "      <td>16</td>\n",
       "      <td>6</td>\n",
       "      <td>2017</td>\n",
       "      <td>0</td>\n",
       "      <td>3.6383</td>\n",
       "      <td>7.4233</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>20080</td>\n",
       "      <td>Montgomery County Police</td>\n",
       "      <td>Injury Crash</td>\n",
       "      <td>County</td>\n",
       "      <td>JANET RD</td>\n",
       "      <td>County</td>\n",
       "      <td>FLACK ST</td>\n",
       "      <td>STRAIGHT MOVEMENT ANGLE</td>\n",
       "      <td>CLEAR</td>\n",
       "      <td>DRY</td>\n",
       "      <td>...</td>\n",
       "      <td>-77.063227</td>\n",
       "      <td>1</td>\n",
       "      <td>M</td>\n",
       "      <td>22</td>\n",
       "      <td>10</td>\n",
       "      <td>2</td>\n",
       "      <td>2020</td>\n",
       "      <td>5</td>\n",
       "      <td>3.0862</td>\n",
       "      <td>7.3240</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>10103</td>\n",
       "      <td>Montgomery County Police</td>\n",
       "      <td>Property Damage Crash</td>\n",
       "      <td>Maryland (State)</td>\n",
       "      <td>COLESVILLE RD</td>\n",
       "      <td>Maryland (State)</td>\n",
       "      <td>STRUC #15082</td>\n",
       "      <td>SAME DIRECTION RIGHT TURN</td>\n",
       "      <td>CLEAR</td>\n",
       "      <td>DRY</td>\n",
       "      <td>...</td>\n",
       "      <td>-77.032718</td>\n",
       "      <td>0</td>\n",
       "      <td>M</td>\n",
       "      <td>11</td>\n",
       "      <td>8</td>\n",
       "      <td>6</td>\n",
       "      <td>2017</td>\n",
       "      <td>6</td>\n",
       "      <td>3.9583</td>\n",
       "      <td>7.3783</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>10208</td>\n",
       "      <td>Montgomery County Police</td>\n",
       "      <td>Injury Crash</td>\n",
       "      <td>County</td>\n",
       "      <td>PARKLAND DR</td>\n",
       "      <td>County</td>\n",
       "      <td>FRANKFORT DR</td>\n",
       "      <td>OTHER</td>\n",
       "      <td>CLOUDY</td>\n",
       "      <td>DRY</td>\n",
       "      <td>...</td>\n",
       "      <td>-77.086933</td>\n",
       "      <td>1</td>\n",
       "      <td>M</td>\n",
       "      <td>10</td>\n",
       "      <td>7</td>\n",
       "      <td>5</td>\n",
       "      <td>2019</td>\n",
       "      <td>4</td>\n",
       "      <td>3.8400</td>\n",
       "      <td>-70633.0000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>1091</td>\n",
       "      <td>Montgomery County Police</td>\n",
       "      <td>Injury Crash</td>\n",
       "      <td>Maryland (State)</td>\n",
       "      <td>OLD GEORGETOWN RD</td>\n",
       "      <td>County</td>\n",
       "      <td>MCKINLEY ST</td>\n",
       "      <td>SAME DIR REAR END</td>\n",
       "      <td>CLEAR</td>\n",
       "      <td>DRY</td>\n",
       "      <td>...</td>\n",
       "      <td>-77.109168</td>\n",
       "      <td>0</td>\n",
       "      <td>M</td>\n",
       "      <td>2</td>\n",
       "      <td>17</td>\n",
       "      <td>8</td>\n",
       "      <td>2016</td>\n",
       "      <td>1</td>\n",
       "      <td>3.9917</td>\n",
       "      <td>7.0183</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 40 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "  Local Case Number               Agency Name       ACRS Report Type  \\\n",
       "0             10196  Montgomery County Police           Injury Crash   \n",
       "1             20080  Montgomery County Police           Injury Crash   \n",
       "2             10103  Montgomery County Police  Property Damage Crash   \n",
       "3             10208  Montgomery County Police           Injury Crash   \n",
       "4              1091  Montgomery County Police           Injury Crash   \n",
       "\n",
       "         Route Type          Road Name Cross-Street Type Cross-Street Name  \\\n",
       "0  Maryland (State)     OLD HUNDRED RD            County       THURSTON RD   \n",
       "1            County           JANET RD            County          FLACK ST   \n",
       "2  Maryland (State)      COLESVILLE RD  Maryland (State)      STRUC #15082   \n",
       "3            County        PARKLAND DR            County      FRANKFORT DR   \n",
       "4  Maryland (State)  OLD GEORGETOWN RD            County       MCKINLEY ST   \n",
       "\n",
       "              Collision Type Weather Surface Condition  ...  Longitude Fault  \\\n",
       "0             SINGLE VEHICLE   CLEAR               WET  ... -77.344203     1   \n",
       "1    STRAIGHT MOVEMENT ANGLE   CLEAR               DRY  ... -77.063227     1   \n",
       "2  SAME DIRECTION RIGHT TURN   CLEAR               DRY  ... -77.032718     0   \n",
       "3                      OTHER  CLOUDY               DRY  ... -77.086933     1   \n",
       "4          SAME DIR REAR END   CLEAR               DRY  ... -77.109168     0   \n",
       "\n",
       "  Report Category Crash_day Crash_hour Crash_month Crash_year  \\\n",
       "0               M         5         16           6       2017   \n",
       "1               M        22         10           2       2020   \n",
       "2               M        11          8           6       2017   \n",
       "3               M        10          7           5       2019   \n",
       "4               M         2         17           8       2016   \n",
       "\n",
       "  Crash_day_of_week Location_x  Location_y  \n",
       "0                 0     3.6383      7.4233  \n",
       "1                 5     3.0862      7.3240  \n",
       "2                 6     3.9583      7.3783  \n",
       "3                 4     3.8400 -70633.0000  \n",
       "4                 1     3.9917      7.0183  \n",
       "\n",
       "[5 rows x 40 columns]"
      ]
     },
     "execution_count": 22,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "id": "2016be6a",
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Local Case Number                  6453\n",
       "Agency Name                           0\n",
       "ACRS Report Type                      0\n",
       "Route Type                         4968\n",
       "Road Name                          4502\n",
       "Cross-Street Type                  5019\n",
       "Cross-Street Name                  4522\n",
       "Collision Type                      299\n",
       "Weather                            4029\n",
       "Surface Condition                  5879\n",
       "Light                               704\n",
       "Traffic Control                    7774\n",
       "Driver Substance Abuse             9095\n",
       "Injury Severity                       0\n",
       "Drivers License State              2528\n",
       "Vehicle Damage Extent               397\n",
       "Vehicle First Impact Location       222\n",
       "Vehicle Second Impact Location      234\n",
       "Vehicle Body Type                   785\n",
       "Vehicle Movement                    198\n",
       "Vehicle Continuing Dir             1129\n",
       "Vehicle Going Dir                  1113\n",
       "Speed Limit                           0\n",
       "Driverless Vehicle                    0\n",
       "Parked Vehicle                        0\n",
       "Vehicle Year                          0\n",
       "Vehicle Make                          8\n",
       "Vehicle Model                        19\n",
       "Equipment Problems                10666\n",
       "Latitude                              0\n",
       "Longitude                             0\n",
       "Fault                                 0\n",
       "Report Category                       0\n",
       "Crash_day                             0\n",
       "Crash_hour                            0\n",
       "Crash_month                           0\n",
       "Crash_year                            0\n",
       "Crash_day_of_week                     0\n",
       "Location_x                            0\n",
       "Location_y                            0\n",
       "dtype: int64"
      ]
     },
     "execution_count": 23,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.isnull().sum()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "id": "16b1e3ec",
   "metadata": {},
   "outputs": [],
   "source": [
    "# now converting the nan values to unknowns"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "id": "4293a364",
   "metadata": {},
   "outputs": [],
   "source": [
    "df['Agency Name'] = df['Agency Name'].replace('UNKNOWN','Unknown')\n",
    "df['Agency Name'] = df['Agency Name'].fillna('Unknown')\n",
    "\n",
    "df['ACRS Report Type'] = df['ACRS Report Type'].replace('UNKNOWN','Unknown')\n",
    "df['ACRS Report Type'] = df['ACRS Report Type'].fillna('Unknown')\n",
    "\n",
    "df['Route Type'] = df['Route Type'].replace('UNKNOWN','Unknown')\n",
    "df['Route Type'] = df['Route Type'].fillna('Unknown')\n",
    "\n",
    "df['Cross-Street Type'] = df['Cross-Street Type'].replace('UNKNOWN','Unknown')\n",
    "df['Cross-Street Type'] = df['Cross-Street Type'].fillna('Unknown')\n",
    "\n",
    "df['Weather'] = df['Weather'].replace('UNKNOWN','Unknown')\n",
    "df['Weather'] = df['Weather'].fillna('Unknown')\n",
    "\n",
    "df['Surface Condition'] = df['Surface Condition'].replace('UNKNOWN','Unknown')\n",
    "df['Surface Condition'] = df['Surface Condition'].fillna('Unknown')\n",
    "\n",
    "df['Light'] = df['Light'].replace('UNKNOWN','Unknown')\n",
    "df['Light'] = df['Light'].fillna('Unknown')\n",
    "\n",
    "df['Traffic Control'] = df['Traffic Control'].replace('UNKNOWN','Unknown')\n",
    "df['Traffic Control'] = df['Traffic Control'].fillna('Unknown')\n",
    "\n",
    "df['Driver Substance Abuse'] = df['Driver Substance Abuse'].replace('UNKNOWN','Unknown')\n",
    "df['Driver Substance Abuse'] = df['Driver Substance Abuse'].fillna('Unknown')\n",
    "\n",
    "df['Injury Severity'] = df['Injury Severity'].replace('UNKNOWN','Unknown')\n",
    "df['Injury Severity'] = df['Injury Severity'].fillna('Unknown')\n",
    "\n",
    "df['Vehicle Damage Extent'] = df['Vehicle Damage Extent'].replace('UNKNOWN','Unknown')\n",
    "df['Vehicle Damage Extent'] = df['Vehicle Damage Extent'].fillna('Unknown')\n",
    "\n",
    "df['Vehicle Second Impact Location'] = df['Vehicle Second Impact Location'].replace('UNKNOWN','Unknown')\n",
    "df['Vehicle Second Impact Location'] = df['Vehicle Second Impact Location'].fillna('Unknown')\n",
    "\n",
    "df['Vehicle Body Type'] = df['Vehicle Body Type'].replace('UNKNOWN','Unknown')\n",
    "df['Vehicle Body Type'] = df['Vehicle Body Type'].fillna('Unknown')\n",
    "\n",
    "df['Vehicle Movement'] = df['Vehicle Movement'].replace('UNKNOWN','Unknown')\n",
    "df['Vehicle Movement'] = df['Vehicle Movement'].fillna('Unknown')\n",
    "\n",
    "df['Vehicle Continuing Dir'] = df['Vehicle Continuing Dir'].replace('UNKNOWN','Unknown')\n",
    "df['Vehicle Continuing Dir'] = df['Vehicle Continuing Dir'].fillna('Unknown')\n",
    "\n",
    "df['Vehicle Going Dir'] = df['Vehicle Going Dir'].replace('UNKNOWN','Unknown')\n",
    "df['Vehicle Going Dir'] = df['Vehicle Going Dir'].fillna('Unknown')\n",
    "\n",
    "df['Equipment Problems'] = df['Equipment Problems'].replace('UNKNOWN','Unknown')\n",
    "df['Equipment Problems'] = df['Equipment Problems'].fillna('Unknown')\n",
    "\n",
    "df['Vehicle First Impact Location'] = df['Vehicle First Impact Location'].replace('UNKNOWN','Unknown')\n",
    "df['Vehicle First Impact Location'] = df['Vehicle First Impact Location'].fillna('Unknown')\n",
    "\n",
    "df['Collision Type'] = df['Collision Type'].replace('UNKNOWN','Unknown')\n",
    "df['Collision Type'] = df['Collision Type'].fillna('Unknown')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "id": "7d8d2a05",
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Local Case Number                 6453\n",
       "Agency Name                          0\n",
       "ACRS Report Type                     0\n",
       "Route Type                           0\n",
       "Road Name                         4502\n",
       "Cross-Street Type                    0\n",
       "Cross-Street Name                 4522\n",
       "Collision Type                       0\n",
       "Weather                              0\n",
       "Surface Condition                    0\n",
       "Light                                0\n",
       "Traffic Control                      0\n",
       "Driver Substance Abuse               0\n",
       "Injury Severity                      0\n",
       "Drivers License State             2528\n",
       "Vehicle Damage Extent                0\n",
       "Vehicle First Impact Location        0\n",
       "Vehicle Second Impact Location       0\n",
       "Vehicle Body Type                    0\n",
       "Vehicle Movement                     0\n",
       "Vehicle Continuing Dir               0\n",
       "Vehicle Going Dir                    0\n",
       "Speed Limit                          0\n",
       "Driverless Vehicle                   0\n",
       "Parked Vehicle                       0\n",
       "Vehicle Year                         0\n",
       "Vehicle Make                         8\n",
       "Vehicle Model                       19\n",
       "Equipment Problems                   0\n",
       "Latitude                             0\n",
       "Longitude                            0\n",
       "Fault                                0\n",
       "Report Category                      0\n",
       "Crash_day                            0\n",
       "Crash_hour                           0\n",
       "Crash_month                          0\n",
       "Crash_year                           0\n",
       "Crash_day_of_week                    0\n",
       "Location_x                           0\n",
       "Location_y                           0\n",
       "dtype: int64"
      ]
     },
     "execution_count": 26,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.isnull().sum()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "id": "f865ea5b",
   "metadata": {},
   "outputs": [],
   "source": [
    "df.fillna(-1,inplace=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "id": "ad19cb56",
   "metadata": {},
   "outputs": [],
   "source": [
    "df['Local Case Number'] = df['Local Case Number'].str.replace('P','')\n",
    "df['Local Case Number']= pd.to_numeric(df['Local Case Number'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "id": "bf63262f",
   "metadata": {},
   "outputs": [],
   "source": [
    "df['Local Case Number'].fillna(-1, inplace = True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "id": "e0e3aacb",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(51490, 40)"
      ]
     },
     "execution_count": 30,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "id": "74531261",
   "metadata": {},
   "outputs": [],
   "source": [
    "# before running this cell make sure there aren't any nan values \n",
    "\n",
    "enc_RN = (df.groupby('Road Name').size()) / len(df)\n",
    "df['enc_RN'] = df['Road Name'].apply(lambda x : enc_RN[x])\n",
    "\n",
    "enc_CSN = (df.groupby('Cross-Street Name').size()) / len(df)\n",
    "df['enc_CSN'] = df['Cross-Street Name'].apply(lambda x : enc_CSN[x])\n",
    "\n",
    "enc_VMK = (df.groupby('Vehicle Make').size()) / len(df)\n",
    "df['enc_VM'] = df['Vehicle Make'].apply(lambda x : enc_VMK[x])\n",
    "\n",
    "enc_VMM = (df.groupby('Vehicle Model').size()) / len(df)\n",
    "df['enc_VMM'] = df['Vehicle Model'].apply(lambda x : enc_VMM[x])\n",
    "\n",
    "df.drop(['Road Name','Cross-Street Name','Vehicle Make','Vehicle Model'], axis = 1, inplace = True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "id": "4c6e5553",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Local Case Number</th>\n",
       "      <th>Agency Name</th>\n",
       "      <th>ACRS Report Type</th>\n",
       "      <th>Route Type</th>\n",
       "      <th>Cross-Street Type</th>\n",
       "      <th>Collision Type</th>\n",
       "      <th>Weather</th>\n",
       "      <th>Surface Condition</th>\n",
       "      <th>Light</th>\n",
       "      <th>Traffic Control</th>\n",
       "      <th>...</th>\n",
       "      <th>Crash_hour</th>\n",
       "      <th>Crash_month</th>\n",
       "      <th>Crash_year</th>\n",
       "      <th>Crash_day_of_week</th>\n",
       "      <th>Location_x</th>\n",
       "      <th>Location_y</th>\n",
       "      <th>enc_RN</th>\n",
       "      <th>enc_CSN</th>\n",
       "      <th>enc_VM</th>\n",
       "      <th>enc_VMM</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>10196.0</td>\n",
       "      <td>Montgomery County Police</td>\n",
       "      <td>Injury Crash</td>\n",
       "      <td>Maryland (State)</td>\n",
       "      <td>County</td>\n",
       "      <td>SINGLE VEHICLE</td>\n",
       "      <td>CLEAR</td>\n",
       "      <td>WET</td>\n",
       "      <td>DAYLIGHT</td>\n",
       "      <td>NO CONTROLS</td>\n",
       "      <td>...</td>\n",
       "      <td>16</td>\n",
       "      <td>6</td>\n",
       "      <td>2017</td>\n",
       "      <td>0</td>\n",
       "      <td>3.6383</td>\n",
       "      <td>7.4233</td>\n",
       "      <td>0.001398</td>\n",
       "      <td>0.000175</td>\n",
       "      <td>0.010818</td>\n",
       "      <td>0.010080</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>20080.0</td>\n",
       "      <td>Montgomery County Police</td>\n",
       "      <td>Injury Crash</td>\n",
       "      <td>County</td>\n",
       "      <td>County</td>\n",
       "      <td>STRAIGHT MOVEMENT ANGLE</td>\n",
       "      <td>CLEAR</td>\n",
       "      <td>DRY</td>\n",
       "      <td>DAYLIGHT</td>\n",
       "      <td>STOP SIGN</td>\n",
       "      <td>...</td>\n",
       "      <td>10</td>\n",
       "      <td>2</td>\n",
       "      <td>2020</td>\n",
       "      <td>5</td>\n",
       "      <td>3.0862</td>\n",
       "      <td>7.3240</td>\n",
       "      <td>0.000155</td>\n",
       "      <td>0.000136</td>\n",
       "      <td>0.096038</td>\n",
       "      <td>0.000990</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>10103.0</td>\n",
       "      <td>Montgomery County Police</td>\n",
       "      <td>Property Damage Crash</td>\n",
       "      <td>Maryland (State)</td>\n",
       "      <td>Maryland (State)</td>\n",
       "      <td>SAME DIRECTION RIGHT TURN</td>\n",
       "      <td>CLEAR</td>\n",
       "      <td>DRY</td>\n",
       "      <td>Unknown</td>\n",
       "      <td>OTHER</td>\n",
       "      <td>...</td>\n",
       "      <td>8</td>\n",
       "      <td>6</td>\n",
       "      <td>2017</td>\n",
       "      <td>6</td>\n",
       "      <td>3.9583</td>\n",
       "      <td>7.3783</td>\n",
       "      <td>0.018625</td>\n",
       "      <td>0.000058</td>\n",
       "      <td>0.004525</td>\n",
       "      <td>0.001728</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>3 rows × 40 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "   Local Case Number               Agency Name       ACRS Report Type  \\\n",
       "0            10196.0  Montgomery County Police           Injury Crash   \n",
       "1            20080.0  Montgomery County Police           Injury Crash   \n",
       "2            10103.0  Montgomery County Police  Property Damage Crash   \n",
       "\n",
       "         Route Type Cross-Street Type             Collision Type Weather  \\\n",
       "0  Maryland (State)            County             SINGLE VEHICLE   CLEAR   \n",
       "1            County            County    STRAIGHT MOVEMENT ANGLE   CLEAR   \n",
       "2  Maryland (State)  Maryland (State)  SAME DIRECTION RIGHT TURN   CLEAR   \n",
       "\n",
       "  Surface Condition     Light Traffic Control  ... Crash_hour Crash_month  \\\n",
       "0               WET  DAYLIGHT     NO CONTROLS  ...         16           6   \n",
       "1               DRY  DAYLIGHT       STOP SIGN  ...         10           2   \n",
       "2               DRY   Unknown           OTHER  ...          8           6   \n",
       "\n",
       "  Crash_year Crash_day_of_week Location_x Location_y    enc_RN   enc_CSN  \\\n",
       "0       2017                 0     3.6383     7.4233  0.001398  0.000175   \n",
       "1       2020                 5     3.0862     7.3240  0.000155  0.000136   \n",
       "2       2017                 6     3.9583     7.3783  0.018625  0.000058   \n",
       "\n",
       "     enc_VM   enc_VMM  \n",
       "0  0.010818  0.010080  \n",
       "1  0.096038  0.000990  \n",
       "2  0.004525  0.001728  \n",
       "\n",
       "[3 rows x 40 columns]"
      ]
     },
     "execution_count": 32,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.head(3)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "14e169cc",
   "metadata": {},
   "source": [
    "#### Changing the categorical features to numerical"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "id": "ccf65457",
   "metadata": {},
   "outputs": [],
   "source": [
    "df = pd.get_dummies(df,drop_first = True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "id": "e3a86115",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(51490, 286)"
      ]
     },
     "execution_count": 34,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "id": "a0f4a8e6",
   "metadata": {},
   "outputs": [],
   "source": [
    "X = df.drop('Fault',axis = 1)\n",
    "y = df['Fault']\n",
    "X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.30, random_state=101)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f44e1113",
   "metadata": {},
   "source": [
    "### Pre-processing and EDA of the testing data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "id": "04eaec34",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(77235, 42)"
      ]
     },
     "execution_count": 36,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_test = pd.read_csv(\"test_accident.csv\",parse_dates = ['Crash Date/Time'])\n",
    "df_test.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "id": "8527f641",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array(['M', 'D', 'E', 'H'], dtype=object)"
      ]
     },
     "execution_count": 37,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "def extract(data):\n",
    "    return data[:1]\n",
    "\n",
    "df_test['Report Category'] = df_test['Report Number'].apply(extract)\n",
    "\n",
    "df_test['Report Category'].unique()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "id": "b7689a01",
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Id                                    0\n",
       "Report Number                         0\n",
       "Local Case Number                 10073\n",
       "Agency Name                           0\n",
       "ACRS Report Type                      0\n",
       "Crash Date/Time                       0\n",
       "Route Type                         7467\n",
       "Road Name                          7069\n",
       "Cross-Street Type                  7480\n",
       "Cross-Street Name                  7071\n",
       "Off-Road Description              70166\n",
       "Municipality                      68737\n",
       "Related Non-Motorist              74987\n",
       "Collision Type                      245\n",
       "Weather                            5996\n",
       "Surface Condition                  9053\n",
       "Light                               620\n",
       "Traffic Control                   12050\n",
       "Driver Substance Abuse            14083\n",
       "Non-Motorist Substance Abuse      75458\n",
       "Person ID                             0\n",
       "Injury Severity                       0\n",
       "Circumstance                      62690\n",
       "Drivers License State              3920\n",
       "Vehicle ID                            0\n",
       "Vehicle Damage Extent               136\n",
       "Vehicle First Impact Location        81\n",
       "Vehicle Second Impact Location      139\n",
       "Vehicle Body Type                  1107\n",
       "Vehicle Movement                    153\n",
       "Vehicle Continuing Dir             1201\n",
       "Vehicle Going Dir                  1201\n",
       "Speed Limit                           0\n",
       "Driverless Vehicle                    0\n",
       "Parked Vehicle                        0\n",
       "Vehicle Year                          0\n",
       "Vehicle Make                         13\n",
       "Vehicle Model                        34\n",
       "Equipment Problems                16424\n",
       "Latitude                              0\n",
       "Longitude                             0\n",
       "Location                              0\n",
       "Report Category                       0\n",
       "dtype: int64"
      ]
     },
     "execution_count": 38,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_test.isnull().sum()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "id": "931ae413",
   "metadata": {},
   "outputs": [],
   "source": [
    "df_test.drop(['Off-Road Description','Municipality','Related Non-Motorist','Non-Motorist Substance Abuse',\n",
    "         'Circumstance','Person ID','Vehicle ID'], axis = 1, inplace = True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "id": "68e6d24a",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(77235, 36)"
      ]
     },
     "execution_count": 40,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_test.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "id": "dd30fd00",
   "metadata": {},
   "outputs": [],
   "source": [
    "df_test['Crash_day'] = df_test['Crash Date/Time'].dt.day\n",
    "df_test['Crash_hour'] = df_test['Crash Date/Time'].dt.hour\n",
    "df_test['Crash_month'] = df_test['Crash Date/Time'].dt.month\n",
    "df_test['Crash_year'] = df_test['Crash Date/Time'].dt.year\n",
    "df_test['Crash_day_of_week'] = df_test['Crash Date/Time'].dt.dayofweek\n",
    "df_test.drop('Crash Date/Time',axis='columns', inplace=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "id": "8d451959",
   "metadata": {},
   "outputs": [],
   "source": [
    "df_test['Location'] = df_test['Location'].str.replace(',',' ')\n",
    "new = df_test['Location'].str.split(\" \", n = 1, expand = True)\n",
    "\n",
    "df_test['Location_x'] = new[0]\n",
    "df_test['Location_y'] = new[1]\n",
    "df_test['Location_x'] = df_test['Location_x'].astype(float)\n",
    "df_test['Location_y'] = df_test['Location_y'].astype(float)\n",
    "\n",
    "df_test.drop('Location', axis = 1, inplace = True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "id": "292ec4e0",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(77235, 41)"
      ]
     },
     "execution_count": 43,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_test.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "id": "961035c9",
   "metadata": {},
   "outputs": [],
   "source": [
    "drop_columns = ['Report Number']\n",
    "df_test.drop(df_test[drop_columns], axis = 1, inplace = True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "id": "7f5dc4d9",
   "metadata": {},
   "outputs": [],
   "source": [
    "df_test['Local Case Number'] = df_test['Local Case Number'].str.replace('P','')\n",
    "df_test['Local Case Number']= pd.to_numeric(df_test['Local Case Number'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "id": "4948b289",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(77235, 40)"
      ]
     },
     "execution_count": 46,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_test.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "id": "17d0c7ae",
   "metadata": {},
   "outputs": [],
   "source": [
    "df_test['Agency Name'] = df_test['Agency Name'].replace('UNKNOWN','Unknown')\n",
    "df_test['Agency Name'] = df_test['Agency Name'].fillna('Unknown')\n",
    "\n",
    "df_test['ACRS Report Type'] = df_test['ACRS Report Type'].replace('UNKNOWN','Unknown')\n",
    "df_test['ACRS Report Type'] = df_test['ACRS Report Type'].fillna('Unknown')\n",
    "\n",
    "df_test['Route Type'] = df_test['Route Type'].replace('UNKNOWN','Unknown')\n",
    "df_test['Route Type'] = df_test['Route Type'].fillna('Unknown')\n",
    "\n",
    "df_test['Cross-Street Type'] = df_test['Cross-Street Type'].replace('UNKNOWN','Unknown')\n",
    "df_test['Cross-Street Type'] = df_test['Cross-Street Type'].fillna('Unknown')\n",
    "\n",
    "df_test['Weather'] = df_test['Weather'].replace('UNKNOWN','Unknown')\n",
    "df_test['Weather'] = df_test['Weather'].fillna('Unknown')\n",
    "\n",
    "df_test['Surface Condition'] = df_test['Surface Condition'].replace('UNKNOWN','Unknown')\n",
    "df_test['Surface Condition'] = df_test['Surface Condition'].fillna('Unknown')\n",
    "\n",
    "df_test['Light'] = df_test['Light'].replace('UNKNOWN','Unknown')\n",
    "df_test['Light'] = df_test['Light'].fillna('Unknown')\n",
    "\n",
    "df_test['Traffic Control'] = df_test['Traffic Control'].replace('UNKNOWN','Unknown')\n",
    "df_test['Traffic Control'] = df_test['Traffic Control'].fillna('Unknown')\n",
    "\n",
    "df_test['Driver Substance Abuse'] = df_test['Driver Substance Abuse'].replace('UNKNOWN','Unknown')\n",
    "df_test['Driver Substance Abuse'] = df_test['Driver Substance Abuse'].fillna('Unknown')\n",
    "\n",
    "df_test['Injury Severity'] = df_test['Injury Severity'].replace('UNKNOWN','Unknown')\n",
    "df_test['Injury Severity'] = df_test['Injury Severity'].fillna('Unknown')\n",
    "\n",
    "df_test['Vehicle Damage Extent'] = df_test['Vehicle Damage Extent'].replace('UNKNOWN','Unknown')\n",
    "df_test['Vehicle Damage Extent'] = df_test['Vehicle Damage Extent'].fillna('Unknown')\n",
    "\n",
    "df_test['Vehicle Second Impact Location'] = df_test['Vehicle Second Impact Location'].replace('UNKNOWN','Unknown')\n",
    "df_test['Vehicle Second Impact Location'] = df_test['Vehicle Second Impact Location'].fillna('Unknown')\n",
    "\n",
    "df_test['Vehicle Body Type'] = df_test['Vehicle Body Type'].replace('UNKNOWN','Unknown')\n",
    "df_test['Vehicle Body Type'] = df_test['Vehicle Body Type'].fillna('Unknown')\n",
    "\n",
    "df_test['Vehicle Movement'] = df_test['Vehicle Movement'].replace('UNKNOWN','Unknown')\n",
    "df_test['Vehicle Movement'] = df_test['Vehicle Movement'].fillna('Unknown')\n",
    "\n",
    "df_test['Vehicle Continuing Dir'] = df_test['Vehicle Continuing Dir'].replace('UNKNOWN','Unknown')\n",
    "df_test['Vehicle Continuing Dir'] = df_test['Vehicle Continuing Dir'].fillna('Unknown')\n",
    "\n",
    "df_test['Vehicle Going Dir'] = df_test['Vehicle Going Dir'].replace('UNKNOWN','Unknown')\n",
    "df_test['Vehicle Going Dir'] = df_test['Vehicle Going Dir'].fillna('Unknown')\n",
    "\n",
    "df_test['Equipment Problems'] = df_test['Equipment Problems'].replace('UNKNOWN','Unknown')\n",
    "df_test['Equipment Problems'] = df_test['Equipment Problems'].fillna('Unknown')\n",
    "\n",
    "df_test['Vehicle First Impact Location'] = df_test['Vehicle First Impact Location'].replace('UNKNOWN','Unknown')\n",
    "df_test['Vehicle First Impact Location'] = df_test['Vehicle First Impact Location'].fillna('Unknown')\n",
    "\n",
    "df_test['Collision Type'] = df_test['Collision Type'].replace('UNKNOWN','Unknown')\n",
    "df_test['Collision Type'] = df_test['Collision Type'].fillna('Unknown')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "id": "ee33d2d2",
   "metadata": {},
   "outputs": [],
   "source": [
    "df_test.fillna(-1,inplace=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "id": "4f178f10",
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Id                                0\n",
       "Local Case Number                 0\n",
       "Agency Name                       0\n",
       "ACRS Report Type                  0\n",
       "Route Type                        0\n",
       "Road Name                         0\n",
       "Cross-Street Type                 0\n",
       "Cross-Street Name                 0\n",
       "Collision Type                    0\n",
       "Weather                           0\n",
       "Surface Condition                 0\n",
       "Light                             0\n",
       "Traffic Control                   0\n",
       "Driver Substance Abuse            0\n",
       "Injury Severity                   0\n",
       "Drivers License State             0\n",
       "Vehicle Damage Extent             0\n",
       "Vehicle First Impact Location     0\n",
       "Vehicle Second Impact Location    0\n",
       "Vehicle Body Type                 0\n",
       "Vehicle Movement                  0\n",
       "Vehicle Continuing Dir            0\n",
       "Vehicle Going Dir                 0\n",
       "Speed Limit                       0\n",
       "Driverless Vehicle                0\n",
       "Parked Vehicle                    0\n",
       "Vehicle Year                      0\n",
       "Vehicle Make                      0\n",
       "Vehicle Model                     0\n",
       "Equipment Problems                0\n",
       "Latitude                          0\n",
       "Longitude                         0\n",
       "Report Category                   0\n",
       "Crash_day                         0\n",
       "Crash_hour                        0\n",
       "Crash_month                       0\n",
       "Crash_year                        0\n",
       "Crash_day_of_week                 0\n",
       "Location_x                        0\n",
       "Location_y                        0\n",
       "dtype: int64"
      ]
     },
     "execution_count": 49,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_test.isnull().sum()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "id": "4b5c870f",
   "metadata": {},
   "outputs": [],
   "source": [
    "# before running this cell make sure there aren't any nan values \n",
    "\n",
    "enc_RN = (df_test.groupby('Road Name').size()) / len(df_test)\n",
    "df_test['enc_RN'] = df_test['Road Name'].apply(lambda x : enc_RN[x])\n",
    "\n",
    "enc_CSN = (df_test.groupby('Cross-Street Name').size()) / len(df_test)\n",
    "df_test['enc_CSN'] = df_test['Cross-Street Name'].apply(lambda x : enc_CSN[x])\n",
    "\n",
    "enc_VMK = (df_test.groupby('Vehicle Make').size()) / len(df_test)\n",
    "df_test['enc_VM'] = df_test['Vehicle Make'].apply(lambda x : enc_VMK[x])\n",
    "\n",
    "enc_VMM = (df_test.groupby('Vehicle Model').size()) / len(df_test)\n",
    "df_test['enc_VMM'] = df_test['Vehicle Model'].apply(lambda x : enc_VMM[x])\n",
    "\n",
    "df_test.drop(['Road Name','Cross-Street Name','Vehicle Make','Vehicle Model'], axis = 1, inplace = True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "id": "f36009e8",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(77235, 40)"
      ]
     },
     "execution_count": 51,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_test.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "e8cf89bd",
   "metadata": {},
   "source": [
    "#### Changing the categorical features to numerical"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "id": "48fff6c3",
   "metadata": {},
   "outputs": [],
   "source": [
    "df_test = pd.get_dummies(df_test,drop_first = True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "id": "318d9c52",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(77235, 296)"
      ]
     },
     "execution_count": 53,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_test.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "id": "e7abe40d",
   "metadata": {},
   "outputs": [],
   "source": [
    "df_test.drop('Id', axis = 1, inplace = True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "id": "467ebd7f",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(77235, 295)"
      ]
     },
     "execution_count": 55,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_test.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "id": "31153d75",
   "metadata": {},
   "outputs": [],
   "source": [
    "df_test.drop('Equipment Problems_SIZE/TYPE IMPROPER',axis=1,inplace=True)\n",
    "df_test.drop('Equipment Problems_NOT STREPPED RIGHT',axis=1,inplace=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "4a3151b6",
   "metadata": {},
   "source": [
    "## Checking the accuracy on the Validation set(30% of training data)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "99363e0d",
   "metadata": {},
   "source": [
    "### Random Forest"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "id": "7c8c9e74",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.849614811937593"
      ]
     },
     "execution_count": 57,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model = RandomForestClassifier()\n",
    "\n",
    "model.fit(X_train, y_train)\n",
    "pred = model.predict(X_test)\n",
    "accuracy_score(y_test,pred)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "dd9521a2",
   "metadata": {},
   "source": [
    "#### Random Forest is giving a validation accuracy of 84.9%"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "b8ad5552",
   "metadata": {},
   "source": [
    "### Gradient Boosting Classifier"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "id": "6f9955ea",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.8472195248268272"
      ]
     },
     "execution_count": 58,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model = GradientBoostingClassifier()\n",
    "\n",
    "model.fit(X_train, y_train)\n",
    "pred = model.predict(X_test)\n",
    "accuracy_score(y_test,pred)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "2a83b5aa",
   "metadata": {},
   "source": [
    "#### Gradient Boost Classifier is giving a validation accuracy of 84.72%"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "134457ff",
   "metadata": {},
   "source": [
    "### XGBoost Classifier"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 59,
   "id": "f41f23da",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.8521395740273192"
      ]
     },
     "execution_count": 59,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model = xgb.XGBClassifier(eval_metric = 'logloss')\n",
    "\n",
    "model.fit(X_train, y_train)\n",
    "pred = model.predict(X_test)\n",
    "accuracy_score(y_test,pred)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f6c4bc7f",
   "metadata": {},
   "source": [
    "#### XGBoost Classifier is giving a validation accuracy of 85.21%"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "42745943",
   "metadata": {},
   "source": [
    "### As XGBoost Classifier is giving the highest validation accuracy out of the rest we will continue with XGBoost Classifier for the hyperparameter tuning"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "12912077",
   "metadata": {},
   "source": [
    "#### Plotting Max Depth Vs Error "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "id": "b40156cc",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "errors=[]\n",
    "range_n = [3,5,7,8,9,10,12,15]\n",
    "\n",
    "for n in range_n:\n",
    "    model = xgb.XGBClassifier(booster = 'dart', n_estimators = 95 , min_child_weight = 3, max_depth = n, learning_rate = 0.1, \n",
    "                          gamma = 0.5, colsample_bytree = 0.4, eta = 0.1, reg_lambda = 0.2,reg_alpha = 0, scale_pos_weight = 1,\n",
    "                         eval_metric = 'logloss')\n",
    "    model.fit(X_train, y_train)\n",
    "    pred = model.predict(X_test)\n",
    "    errors.append(1-accuracy_score(y_test,pred))\n",
    "    \n",
    "plt.plot(range_n,errors)\n",
    "plt.xlabel('Max Depth')\n",
    "plt.ylabel('Error')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "4906ab0f",
   "metadata": {},
   "source": [
    "#### Plotting Learning Rate Vs Error "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 61,
   "id": "453f18f4",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "errors=[]\n",
    "range_n = [0.05,0.07,0.08,0.1,0.12,0.15,0.18,0.25]\n",
    "\n",
    "for n in range_n:\n",
    "    model = xgb.XGBClassifier(booster = 'dart', n_estimators = 95 , min_child_weight = 3, max_depth = 8, learning_rate = n, \n",
    "                          gamma = 0.5, colsample_bytree = 0.4, eta = 0.1, reg_lambda = 0.2,reg_alpha = 0, scale_pos_weight = 1,\n",
    "                         eval_metric = 'logloss')\n",
    "    model.fit(X_train, y_train)\n",
    "    pred = model.predict(X_test)\n",
    "    errors.append(1-accuracy_score(y_test,pred))\n",
    "    \n",
    "plt.plot(range_n,errors)\n",
    "plt.xlabel('Learning Rate')\n",
    "plt.ylabel('Error')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "d677de27",
   "metadata": {},
   "source": [
    "#### Plotting Number of Estimators Vs Error"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 62,
   "id": "ab540034",
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "errors=[]\n",
    "range_n = [40,50,60,70,80,90,100,110,120,130,140,150,160,170,180]\n",
    "\n",
    "for n in range_n:\n",
    "    model = xgb.XGBClassifier(booster = 'dart', n_estimators = n , min_child_weight = 3, max_depth = 8, learning_rate = 0.1, \n",
    "                          gamma = 0.5, colsample_bytree = 0.4, eta = 0.1, reg_lambda = 0.2,reg_alpha = 0, scale_pos_weight = 1,\n",
    "                         eval_metric = 'logloss')\n",
    "    model.fit(X_train, y_train)\n",
    "    pred = model.predict(X_test)\n",
    "    errors.append(1-accuracy_score(y_test,pred))\n",
    "\n",
    "plt.plot(range_n,errors)\n",
    "plt.xlabel('No. of estimators')\n",
    "plt.ylabel('Error')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f3864bbf",
   "metadata": {},
   "source": [
    "#### Checking Validation Accuracy after Hyperparameter Tuning"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 63,
   "id": "b59ee57e",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.8579012105910533"
      ]
     },
     "execution_count": 63,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model = xgb.XGBClassifier(booster = 'dart', n_estimators = 120 ,min_child_weight = 3, max_depth = 9, learning_rate = 0.1, \n",
    "                          gamma = 0.5, colsample_bytree = 0.4, eta = 0.1, reg_lambda = 0.2,reg_alpha = 0, scale_pos_weight = 1,\n",
    "                         eval_metric = 'logloss')\n",
    "\n",
    "model.fit(X_train, y_train)\n",
    "pred = model.predict(X_test)\n",
    "accuracy_score(y_test,pred)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "2a246cc4",
   "metadata": {},
   "source": [
    "### We get a Validation Accuracy of 85.8%"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "109ced6c",
   "metadata": {},
   "source": [
    "### Ploting the Feature Importance Graph"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 64,
   "id": "0e92571e",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1000x400 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "imp = pd.DataFrame(index = X.columns, data = model.feature_importances_, columns = ['Importance'])\n",
    "imp = imp[imp['Importance'] > 0.01]\n",
    "\n",
    "plt.figure(figsize=(10,4),dpi=100)\n",
    "sns.barplot(data = imp.sort_values('Importance'), x = imp.sort_values('Importance').index, y = 'Importance')\n",
    "plt.xticks(rotation=90)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 65,
   "id": "f1007091",
   "metadata": {},
   "outputs": [],
   "source": [
    "y = df['Fault']\n",
    "\n",
    "a = (np.intersect1d(df.columns, df_test.columns))\n",
    "\n",
    "df_test = df_test.loc[:,a]\n",
    "df= df.loc[:,a]\n",
    "\n",
    "X = df"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a979b630",
   "metadata": {},
   "source": [
    "## Model is ready to be tested on the test data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 66,
   "id": "9c621cc2",
   "metadata": {},
   "outputs": [],
   "source": [
    "model = xgb.XGBClassifier(booster = 'dart', n_estimators = 150, min_child_weight = 3, max_depth = 9, learning_rate = 0.1, \n",
    "                          gamma = 0.5, colsample_bytree = 0.4, eta = 0.1, reg_lambda = 0.2,reg_alpha = 0, scale_pos_weight = 1,\n",
    "                         eval_metric = 'logloss')\n",
    "model.fit(X, y)\n",
    "pred = model.predict(df_test)\n",
    "predicted_df = pd.DataFrame({'Fault': pd.Series([], dtype='int')})\n",
    "\n",
    "predicted_df['Fault'] = pred\n",
    "predicted_df = predicted_df.reset_index()\n",
    "predicted_df.columns = ['Id','Fault']\n",
    "predicted_df.to_csv(\"submission.csv\",index = False)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "b0bbdf90",
   "metadata": {},
   "source": [
    "### Achieved a Testing Accuracy of 87.1% on Kaggle using a single model"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "e973e50b",
   "metadata": {},
   "source": [
    "## Running 5 models with different hyperparameters"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 67,
   "id": "61b8ed6e",
   "metadata": {},
   "outputs": [],
   "source": [
    "model = xgb.XGBClassifier(booster = 'dart', n_estimators = 150, min_child_weight = 3, max_depth = 9, learning_rate = 0.1, \n",
    "                          gamma = 0.5, colsample_bytree = 0.4, eta = 0.1, reg_lambda = 0.2,reg_alpha = 0, scale_pos_weight = 1,\n",
    "                         eval_metric = 'logloss')\n",
    "model.fit(X, y)\n",
    "pred = model.predict(df_test)\n",
    "\n",
    "predicted_df = pd.DataFrame({'Fault': pd.Series([], dtype='int')})\n",
    "\n",
    "predicted_df['Fault'] = pred\n",
    "# predicted_df.to_csv(\"hybrid1.csv\")\n",
    "\n",
    "#-------------------------------------------------------------------------------------------------------------------------\n",
    "\n",
    "model = xgb.XGBClassifier(booster = 'dart', n_estimators = 150, min_child_weight = 3, max_depth = 8, learning_rate = 0.3, \n",
    "                          gamma = 0.5, colsample_bytree = 0.4, eta = 0.1, reg_lambda = 0.2,reg_alpha = 0, scale_pos_weight = 1,\n",
    "                         eval_metric = 'logloss')\n",
    "model.fit(X, y)\n",
    "pred = model.predict(df_test)\n",
    "\n",
    "predicted_df = pd.DataFrame({'Fault': pd.Series([], dtype='int')})\n",
    "\n",
    "predicted_df['Fault'] = pred\n",
    "# predicted_df.to_csv(\"hybrid2.csv\")\n",
    "\n",
    "#-------------------------------------------------------------------------------------------------------------------------\n",
    "\n",
    "model = xgb.XGBClassifier(booster = 'dart', n_estimators = 120, min_child_weight = 3, max_depth = 8, learning_rate = 0.4, \n",
    "                          gamma = 0.5, colsample_bytree = 0.4, eta = 0.1, reg_lambda = 0.2,reg_alpha = 0, scale_pos_weight = 1,\n",
    "                         eval_metric = 'logloss')\n",
    "model.fit(X, y)\n",
    "pred = model.predict(df_test)\n",
    "\n",
    "predicted_df = pd.DataFrame({'Fault': pd.Series([], dtype='int')})\n",
    "\n",
    "predicted_df['Fault'] = pred\n",
    "# predicted_df.to_csv(\"hybrid3.csv\")\n",
    "\n",
    "#-------------------------------------------------------------------------------------------------------------------------\n",
    "\n",
    "model = xgb.XGBClassifier(booster = 'dart', n_estimators = 180, min_child_weight = 3, max_depth = 9, learning_rate = 0.4, \n",
    "                          gamma = 0.5, colsample_bytree = 0.4, eta = 0.1, reg_lambda = 0.2,reg_alpha = 0, scale_pos_weight = 1,\n",
    "                         eval_metric = 'logloss')\n",
    "model.fit(X, y)\n",
    "pred = model.predict(df_test)\n",
    "\n",
    "predicted_df = pd.DataFrame({'Fault': pd.Series([], dtype='int')})\n",
    "\n",
    "predicted_df['Fault'] = pred\n",
    "# predicted_df.to_csv(\"hybrid4.csv\")\n",
    "\n",
    "#-------------------------------------------------------------------------------------------------------------------------\n",
    "\n",
    "model = xgb.XGBClassifier(booster = 'dart', n_estimators = 80, min_child_weight = 3, max_depth = 9, learning_rate = 0.3, \n",
    "                          gamma = 0.5, colsample_bytree = 0.4, eta = 0.1, reg_lambda = 0.2,reg_alpha = 0, scale_pos_weight = 1,\n",
    "                         eval_metric = 'logloss')\n",
    "model.fit(X, y)\n",
    "pred = model.predict(df_test)\n",
    "\n",
    "predicted_df = pd.DataFrame({'Fault': pd.Series([], dtype='int')})\n",
    "\n",
    "predicted_df['Fault'] = pred\n",
    "# predicted_df.to_csv(\"hybrid5.csv\")\n",
    "\n",
    "#-------------------------------------------------------------------------------------------------------------------------"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "c4f21ca4",
   "metadata": {},
   "source": [
    "## In order to Push the Accuracy even higher and to get to the top of the leaderboards, we have taken 5 of our best accuracy models and did a majority voting(similar idea as random forests)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 68,
   "id": "d7e2ba8c",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Id</th>\n",
       "      <th>Fault</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>3</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>4</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>77230</th>\n",
       "      <td>77230</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>77231</th>\n",
       "      <td>77231</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>77232</th>\n",
       "      <td>77232</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>77233</th>\n",
       "      <td>77233</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>77234</th>\n",
       "      <td>77234</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>77235 rows × 2 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "          Id  Fault\n",
       "0          0      1\n",
       "1          1      0\n",
       "2          2      1\n",
       "3          3      1\n",
       "4          4      1\n",
       "...      ...    ...\n",
       "77230  77230      0\n",
       "77231  77231      0\n",
       "77232  77232      0\n",
       "77233  77233      1\n",
       "77234  77234      1\n",
       "\n",
       "[77235 rows x 2 columns]"
      ]
     },
     "execution_count": 68,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "sub1 = pd.read_csv('Top 5 CSV/1.csv')\n",
    "sub_f1 = sub1['Fault']\n",
    "\n",
    "sub2 = pd.read_csv('Top 5 CSV/2.csv')\n",
    "sub_f2 = sub2['Fault']\n",
    "\n",
    "sub3 = pd.read_csv('Top 5 CSV/3.csv')\n",
    "sub_f3 = sub3['Fault']\n",
    "\n",
    "sub4 = pd.read_csv('Top 5 CSV/4.csv')\n",
    "sub_f4 = sub4['Fault']\n",
    "\n",
    "sub5 = pd.read_csv('Top 5 CSV/5.csv')\n",
    "sub_f5 = sub5['Fault']\n",
    "\n",
    "\n",
    "final = sub_f1 + sub_f2 + sub_f3 + sub_f4 + sub_f5 \n",
    "final = final/5\n",
    "final = final > 0.5\n",
    "final = np.array(final, dtype = 'int')\n",
    "\n",
    "predicted_df = pd.DataFrame({'Fault': pd.Series([], dtype='int')})\n",
    "\n",
    "predicted_df['Fault'] = final\n",
    "predicted_df = predicted_df.reset_index()\n",
    "predicted_df.columns = ['Id','Fault']\n",
    "predicted_df.to_csv(\"hybrid_submission.csv\",index = False)\n",
    "new = pd.read_csv('hybrid_submission.csv')\n",
    "new"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "789f5e3a",
   "metadata": {},
   "source": [
    "### By majority voting of 5 of our best models we have achieved an accuracy of 88.31% on the Test data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "550b8750",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.8.8"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}