Analysis&Predictive Modelling of Proj.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "ceea86b0",
   "metadata": {},
   "source": [
    "<img src=\"logo_img/logo.png\" height=50 width=50 align=\"right\"/><br/><br/>\n",
    "\n",
    "\n",
    "\n",
    "# IST 5520: Data Science and ML in Python\n",
    "## Project Report: Group 4 \n",
    "\n",
    "<h3>Group Members:&nbsp;<font size=3><i>Sayantan Majumdar, Dawit Wolday Asfaw, Cong Shen, Divya Reddy Manku, Akhil Reddy Annreddy </i></font></h3> \n",
    "\n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "bb67dc8a",
   "metadata": {},
   "source": [
    "# 1. Introduction\n",
    "\n",
    "## 1.1 Background\n",
    "Greenhouse gases (GHG) are responsible for trapping heat, thereby making the planet warmer. For the past 150 years, the sharp global increase in GHG in the atmosphere can be primarily attributed to anthropogenic or human-induced activities. In the United States, the largest source of GHG emissions from human activities includes fossil fuel burning, which acts as the predominant energy source in the transportation sector and is also used to generate electricity and heat. The transportation sector generates the largest share of GHG emissions (29% of 2019 GHG emissions) from burning fossil fuels (mostly petroleum by-products such as gasoline and diesel) for our cars, trucks, ships, trains, and planes.\n",
    "\n",
    "\n",
    "The Environmental Protection Agency (EPA) emissions score (smog rating) reflects vehicle tailpipe emissions (CO$_2$) that contribute to local and regional air pollution, creating problems such as smog, haze, and health issues. CO$_2$ emissions typically constitute 99% of the tailpipe emissions of greenhouse gases. Essentially, the higher the rating, the cleaner the car is, and this rating is a direct proxy for the CO$_2$ emissions.  \n",
    "\n",
    "## 1.2 Problem Description and Research Questions\n",
    "This project aims to build a machine learning model to predict the EPA emission score from different vehicles operating in the United States. Essentially, this is a classification problem wherein we will utilize the EPA vehicle and emission data sets to predict the EPA emission score (a discrete quantity). We will also be analyzing the factors having the most impact on this score which varies from 1 (worst) to 10 (best). In addition, we will compare different machine learning models to assess their performance on the data sets. Such kind of modeling and analysis, if pushed to a production environment, could enable policymakers to address critical issues in the sustainable transportation industry. In this regard, we have to answer the following research question:\n",
    "\n",
    "<i> Which predictors are most suited for predicting the EPA emission score? </i>\n",
    "\n",
    "\n",
    "Some specific research questions include:\n",
    "\n",
    "<i>a.  Which machine learning model works best for this data set?</i> \n",
    "\n",
    "<i>b. Which vehicle class has poor emission ratings? </i>\n",
    "\n",
    "\n",
    "## 1.3 Classification Analysis\n",
    "The vehicle and car emissions data sets can model a classification problem where we need to predict the EPA emission score (a unitless quantity having integral values in the [1,10] interval) based on the explanatory variables. In this project, we deal with a high-dimensional classification problem concerned with prediction (discrete EPA emission score) and inference.\n",
    "\n",
    "## 1.4 Potential Problems and Challenges\n",
    "The major challenge in this project is to perform feature engineering. Appropriate data-preprocessing and feature engineering must be performed to reduce the data dimensionality required to address the above research questions suitably.\n",
    "\n",
    "## 1.5 Tentative Timeline\n",
    "\n",
    "|Phase|Activities|Completion|\n",
    "|----|----------|--------|\n",
    "|Data Collection| Collecting the required data set for the business analytics project|September 24, 2021|\n",
    "|Kickoff|Understanding the project requirements and elicitation|October 1, 2021|\n",
    "|Data Management|Data cleaning, pre-processing|October 29, 2021|\n",
    "|Full Data Analysis|Evaluating different ML models|November 20, 2021|\n",
    "|Project Submission|Report writing, preparing presentation, and proof-reading|December 5, 2021|\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "0f48094a",
   "metadata": {},
   "source": [
    "# 2. Data Source and Collection\n",
    "\n",
    "## 2.1 Data Source\n",
    "\n",
    "This dataset can be found at https://www.fueleconomy.gov/feg/ws/index.shtml. The vehicle and emission data sets need to be linked based on the vehicle ID. The vehicle data set contains 83 explanatory variables (columns) that provide detailed car specifications and has 44075 rows. As for the emissions data set, it has 8 features and 42442 rows. Some of the important features are listed below:\n",
    "\n",
    "|Data Set|Feature Name|Details|\n",
    "|----|----------|--------|\n",
    "|emissions|score|EPA 1-10 smog rating for fuelType1 (target variable)|\n",
    "|vehicle|fuelType1|For single fuel vehicles, this will be the only fuel. For dual fuel vehicles, this will be the conventional fuel.|\n",
    "|vehicle|highway08|Highway MPG for fuelType1|\n",
    "|vehicle|barrels08|Annual petroleum consumption in barrels for fuelType1|\n",
    "|vehicle|year|Model year|\n",
    "|vehicle|VClass|EPA vehicle size class|\n",
    "|vehicle|phevBlended|If True, this vehicle operates on a blend of gasoline and electricity in charge depleting mode|\n",
    "\n",
    "\n",
    "\n",
    "## 2.2 Collection\n",
    "\n",
    "We found this data set on [EPA fuel economy portal](https://www.fueleconomy.gov/). The EPA has generated [annual reports](https://www.fueleconomy.gov/feg/pdfs/guides/FEG2021.pdf), but there are no existing publicly available notebooks having detailed machine learning workflows. Moreover, in conjunction, these two data sets satisfy all the other project requirements ($\\ge$20 columns and $\\ge$1000 rows) listed as part of this group assignment.  \n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "3f859659",
   "metadata": {},
   "source": [
    "# 3. Data Management"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 135,
   "id": "baeff422",
   "metadata": {},
   "outputs": [],
   "source": [
    "# import required libraries for data management\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "import seaborn as sns\n",
    "import plotly.express as px\n",
    "import matplotlib.pyplot as plt\n",
    "from sklearn.preprocessing import RobustScaler\n",
    "from sklearn.decomposition import PCA"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a02cad77",
   "metadata": {},
   "source": [
    "## 3.1 Data Cleaning"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 136,
   "id": "6ded25f1",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Read data sets\n",
    "vehicle_df = pd.read_csv('data/vehicles.csv', low_memory=False)\n",
    "emissions_df = pd.read_csv('data/emissions.csv')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 137,
   "id": "b30b6c9b",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "RangeIndex: 44075 entries, 0 to 44074\n",
      "Data columns (total 83 columns):\n",
      " #   Column           Non-Null Count  Dtype  \n",
      "---  ------           --------------  -----  \n",
      " 0   barrels08        44075 non-null  float64\n",
      " 1   barrelsA08       44075 non-null  float64\n",
      " 2   charge120        44075 non-null  float64\n",
      " 3   charge240        44075 non-null  float64\n",
      " 4   city08           44075 non-null  int64  \n",
      " 5   city08U          44075 non-null  float64\n",
      " 6   cityA08          44075 non-null  int64  \n",
      " 7   cityA08U         44075 non-null  float64\n",
      " 8   cityCD           44075 non-null  float64\n",
      " 9   cityE            44075 non-null  float64\n",
      " 10  cityUF           44075 non-null  float64\n",
      " 11  co2              44075 non-null  int64  \n",
      " 12  co2A             44075 non-null  int64  \n",
      " 13  co2TailpipeAGpm  44075 non-null  float64\n",
      " 14  co2TailpipeGpm   44075 non-null  float64\n",
      " 15  comb08           44075 non-null  int64  \n",
      " 16  comb08U          44075 non-null  float64\n",
      " 17  combA08          44075 non-null  int64  \n",
      " 18  combA08U         44075 non-null  float64\n",
      " 19  combE            44075 non-null  float64\n",
      " 20  combinedCD       44075 non-null  float64\n",
      " 21  combinedUF       44075 non-null  float64\n",
      " 22  cylinders        43770 non-null  float64\n",
      " 23  displ            43772 non-null  float64\n",
      " 24  drive            42889 non-null  object \n",
      " 25  engId            44075 non-null  int64  \n",
      " 26  eng_dscr         27463 non-null  object \n",
      " 27  feScore          44075 non-null  int64  \n",
      " 28  fuelCost08       44075 non-null  int64  \n",
      " 29  fuelCostA08      44075 non-null  int64  \n",
      " 30  fuelType         44075 non-null  object \n",
      " 31  fuelType1        44075 non-null  object \n",
      " 32  ghgScore         44075 non-null  int64  \n",
      " 33  ghgScoreA        44075 non-null  int64  \n",
      " 34  highway08        44075 non-null  int64  \n",
      " 35  highway08U       44075 non-null  float64\n",
      " 36  highwayA08       44075 non-null  int64  \n",
      " 37  highwayA08U      44075 non-null  float64\n",
      " 38  highwayCD        44075 non-null  float64\n",
      " 39  highwayE         44075 non-null  float64\n",
      " 40  highwayUF        44075 non-null  float64\n",
      " 41  hlv              44075 non-null  int64  \n",
      " 42  hpv              44075 non-null  int64  \n",
      " 43  id               44075 non-null  int64  \n",
      " 44  lv2              44075 non-null  int64  \n",
      " 45  lv4              44075 non-null  int64  \n",
      " 46  make             44075 non-null  object \n",
      " 47  model            44075 non-null  object \n",
      " 48  mpgData          44040 non-null  object \n",
      " 49  phevBlended      44075 non-null  bool   \n",
      " 50  pv2              44075 non-null  int64  \n",
      " 51  pv4              44075 non-null  int64  \n",
      " 52  range            44075 non-null  int64  \n",
      " 53  rangeCity        44075 non-null  float64\n",
      " 54  rangeCityA       44075 non-null  float64\n",
      " 55  rangeHwy         44075 non-null  float64\n",
      " 56  rangeHwyA        44075 non-null  float64\n",
      " 57  trany            44064 non-null  object \n",
      " 58  UCity            44075 non-null  float64\n",
      " 59  UCityA           44075 non-null  float64\n",
      " 60  UHighway         44075 non-null  float64\n",
      " 61  UHighwayA        44075 non-null  float64\n",
      " 62  VClass           44075 non-null  object \n",
      " 63  year             44075 non-null  int64  \n",
      " 64  youSaveSpend     44075 non-null  int64  \n",
      " 65  guzzler          2597 non-null   object \n",
      " 66  trans_dscr       15044 non-null  object \n",
      " 67  tCharger         8514 non-null   object \n",
      " 68  sCharger         952 non-null    object \n",
      " 69  atvType          4141 non-null   object \n",
      " 70  fuelType2        1764 non-null   object \n",
      " 71  rangeA           1759 non-null   object \n",
      " 72  evMotor          1363 non-null   object \n",
      " 73  mfrCode          13267 non-null  object \n",
      " 74  c240Dscr         118 non-null    object \n",
      " 75  charge240b       44075 non-null  float64\n",
      " 76  c240bDscr        112 non-null    object \n",
      " 77  createdOn        44075 non-null  object \n",
      " 78  modifiedOn       44075 non-null  object \n",
      " 79  startStop        12386 non-null  object \n",
      " 80  phevCity         44075 non-null  int64  \n",
      " 81  phevHwy          44075 non-null  int64  \n",
      " 82  phevComb         44075 non-null  int64  \n",
      "dtypes: bool(1), float64(32), int64(27), object(23)\n",
      "memory usage: 27.6+ MB\n"
     ]
    },
    {
     "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>count</th>\n",
       "      <th>mean</th>\n",
       "      <th>std</th>\n",
       "      <th>min</th>\n",
       "      <th>25%</th>\n",
       "      <th>50%</th>\n",
       "      <th>75%</th>\n",
       "      <th>max</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>barrels08</th>\n",
       "      <td>44075.0</td>\n",
       "      <td>17.089905</td>\n",
       "      <td>4.698712</td>\n",
       "      <td>0.06</td>\n",
       "      <td>14.330870</td>\n",
       "      <td>16.4805</td>\n",
       "      <td>19.388824</td>\n",
       "      <td>47.087143</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>barrelsA08</th>\n",
       "      <td>44075.0</td>\n",
       "      <td>0.222572</td>\n",
       "      <td>1.143200</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>18.311667</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>charge120</th>\n",
       "      <td>44075.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>charge240</th>\n",
       "      <td>44075.0</td>\n",
       "      <td>0.073465</td>\n",
       "      <td>0.784051</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>15.300000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>city08</th>\n",
       "      <td>44075.0</td>\n",
       "      <td>18.749223</td>\n",
       "      <td>9.036866</td>\n",
       "      <td>6.00</td>\n",
       "      <td>15.000000</td>\n",
       "      <td>17.0000</td>\n",
       "      <td>21.000000</td>\n",
       "      <td>150.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>city08U</th>\n",
       "      <td>44075.0</td>\n",
       "      <td>7.185707</td>\n",
       "      <td>12.972132</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0000</td>\n",
       "      <td>16.169200</td>\n",
       "      <td>150.195800</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>cityA08</th>\n",
       "      <td>44075.0</td>\n",
       "      <td>0.814770</td>\n",
       "      <td>6.280906</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>145.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>cityA08U</th>\n",
       "      <td>44075.0</td>\n",
       "      <td>0.679923</td>\n",
       "      <td>6.181746</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>145.083500</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>cityCD</th>\n",
       "      <td>44075.0</td>\n",
       "      <td>0.000501</td>\n",
       "      <td>0.037958</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>5.350000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>cityE</th>\n",
       "      <td>44075.0</td>\n",
       "      <td>0.520147</td>\n",
       "      <td>4.940315</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>122.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>cityUF</th>\n",
       "      <td>44075.0</td>\n",
       "      <td>0.002852</td>\n",
       "      <td>0.039264</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.927000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>co2</th>\n",
       "      <td>44075.0</td>\n",
       "      <td>107.957686</td>\n",
       "      <td>187.773444</td>\n",
       "      <td>-1.00</td>\n",
       "      <td>-1.000000</td>\n",
       "      <td>-1.0000</td>\n",
       "      <td>275.000000</td>\n",
       "      <td>893.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>co2A</th>\n",
       "      <td>44075.0</td>\n",
       "      <td>5.843607</td>\n",
       "      <td>57.032958</td>\n",
       "      <td>-1.00</td>\n",
       "      <td>-1.000000</td>\n",
       "      <td>-1.0000</td>\n",
       "      <td>-1.000000</td>\n",
       "      <td>713.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>co2TailpipeAGpm</th>\n",
       "      <td>44075.0</td>\n",
       "      <td>16.994214</td>\n",
       "      <td>91.955091</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>713.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>co2TailpipeGpm</th>\n",
       "      <td>44075.0</td>\n",
       "      <td>461.017357</td>\n",
       "      <td>125.764745</td>\n",
       "      <td>0.00</td>\n",
       "      <td>386.391304</td>\n",
       "      <td>444.3500</td>\n",
       "      <td>522.764706</td>\n",
       "      <td>1269.571429</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>comb08</th>\n",
       "      <td>44075.0</td>\n",
       "      <td>20.985071</td>\n",
       "      <td>8.652291</td>\n",
       "      <td>7.00</td>\n",
       "      <td>17.000000</td>\n",
       "      <td>20.0000</td>\n",
       "      <td>23.000000</td>\n",
       "      <td>142.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>comb08U</th>\n",
       "      <td>44075.0</td>\n",
       "      <td>7.972696</td>\n",
       "      <td>13.548087</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0000</td>\n",
       "      <td>18.632100</td>\n",
       "      <td>141.951200</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>combA08</th>\n",
       "      <td>44075.0</td>\n",
       "      <td>0.869541</td>\n",
       "      <td>6.250612</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>133.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>combA08U</th>\n",
       "      <td>44075.0</td>\n",
       "      <td>0.713475</td>\n",
       "      <td>6.110493</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>133.266200</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>combE</th>\n",
       "      <td>44075.0</td>\n",
       "      <td>0.528292</td>\n",
       "      <td>4.962906</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>121.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>combinedCD</th>\n",
       "      <td>44075.0</td>\n",
       "      <td>0.000389</td>\n",
       "      <td>0.033056</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>4.800000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>combinedUF</th>\n",
       "      <td>44075.0</td>\n",
       "      <td>0.002825</td>\n",
       "      <td>0.038834</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.920000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>cylinders</th>\n",
       "      <td>43770.0</td>\n",
       "      <td>5.708933</td>\n",
       "      <td>1.766317</td>\n",
       "      <td>2.00</td>\n",
       "      <td>4.000000</td>\n",
       "      <td>6.0000</td>\n",
       "      <td>6.000000</td>\n",
       "      <td>16.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>displ</th>\n",
       "      <td>43772.0</td>\n",
       "      <td>3.282066</td>\n",
       "      <td>1.356857</td>\n",
       "      <td>0.00</td>\n",
       "      <td>2.200000</td>\n",
       "      <td>3.0000</td>\n",
       "      <td>4.200000</td>\n",
       "      <td>8.400000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>engId</th>\n",
       "      <td>44075.0</td>\n",
       "      <td>7640.478775</td>\n",
       "      <td>16792.609574</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>162.0000</td>\n",
       "      <td>4117.000000</td>\n",
       "      <td>69102.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>feScore</th>\n",
       "      <td>44075.0</td>\n",
       "      <td>0.656858</td>\n",
       "      <td>2.881343</td>\n",
       "      <td>-1.00</td>\n",
       "      <td>-1.000000</td>\n",
       "      <td>-1.0000</td>\n",
       "      <td>3.000000</td>\n",
       "      <td>10.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>fuelCost08</th>\n",
       "      <td>44075.0</td>\n",
       "      <td>2636.197391</td>\n",
       "      <td>738.408030</td>\n",
       "      <td>450.00</td>\n",
       "      <td>2150.000000</td>\n",
       "      <td>2650.0000</td>\n",
       "      <td>3050.000000</td>\n",
       "      <td>8300.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>fuelCostA08</th>\n",
       "      <td>44075.0</td>\n",
       "      <td>98.241634</td>\n",
       "      <td>525.379257</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>4000.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>ghgScore</th>\n",
       "      <td>44075.0</td>\n",
       "      <td>0.658015</td>\n",
       "      <td>2.887266</td>\n",
       "      <td>-1.00</td>\n",
       "      <td>-1.000000</td>\n",
       "      <td>-1.0000</td>\n",
       "      <td>3.000000</td>\n",
       "      <td>10.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>ghgScoreA</th>\n",
       "      <td>44075.0</td>\n",
       "      <td>-0.922042</td>\n",
       "      <td>0.651828</td>\n",
       "      <td>-1.00</td>\n",
       "      <td>-1.000000</td>\n",
       "      <td>-1.0000</td>\n",
       "      <td>-1.000000</td>\n",
       "      <td>8.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>highway08</th>\n",
       "      <td>44075.0</td>\n",
       "      <td>24.853613</td>\n",
       "      <td>8.484246</td>\n",
       "      <td>9.00</td>\n",
       "      <td>20.000000</td>\n",
       "      <td>24.0000</td>\n",
       "      <td>28.000000</td>\n",
       "      <td>133.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>highway08U</th>\n",
       "      <td>44075.0</td>\n",
       "      <td>9.335541</td>\n",
       "      <td>14.939994</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0000</td>\n",
       "      <td>22.780350</td>\n",
       "      <td>133.026300</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>highwayA08</th>\n",
       "      <td>44075.0</td>\n",
       "      <td>0.967487</td>\n",
       "      <td>6.381416</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>121.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>highwayA08U</th>\n",
       "      <td>44075.0</td>\n",
       "      <td>0.779874</td>\n",
       "      <td>6.178836</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>121.200500</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>highwayCD</th>\n",
       "      <td>44075.0</td>\n",
       "      <td>0.000250</td>\n",
       "      <td>0.027495</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>4.060000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>highwayE</th>\n",
       "      <td>44075.0</td>\n",
       "      <td>0.538645</td>\n",
       "      <td>5.012180</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>120.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>highwayUF</th>\n",
       "      <td>44075.0</td>\n",
       "      <td>0.002788</td>\n",
       "      <td>0.038284</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.910000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>hlv</th>\n",
       "      <td>44075.0</td>\n",
       "      <td>1.975201</td>\n",
       "      <td>5.912968</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>49.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>hpv</th>\n",
       "      <td>44075.0</td>\n",
       "      <td>10.052706</td>\n",
       "      <td>27.812047</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>195.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>id</th>\n",
       "      <td>44075.0</td>\n",
       "      <td>22179.946001</td>\n",
       "      <td>12869.366833</td>\n",
       "      <td>1.00</td>\n",
       "      <td>11019.500000</td>\n",
       "      <td>22042.0000</td>\n",
       "      <td>33359.500000</td>\n",
       "      <td>44497.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>lv2</th>\n",
       "      <td>44075.0</td>\n",
       "      <td>1.769212</td>\n",
       "      <td>4.314686</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>41.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>lv4</th>\n",
       "      <td>44075.0</td>\n",
       "      <td>6.034532</td>\n",
       "      <td>9.557379</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0000</td>\n",
       "      <td>13.000000</td>\n",
       "      <td>55.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>pv2</th>\n",
       "      <td>44075.0</td>\n",
       "      <td>13.315553</td>\n",
       "      <td>30.862476</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>194.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>pv4</th>\n",
       "      <td>44075.0</td>\n",
       "      <td>33.536948</td>\n",
       "      <td>46.017789</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0000</td>\n",
       "      <td>91.000000</td>\n",
       "      <td>192.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>range</th>\n",
       "      <td>44075.0</td>\n",
       "      <td>1.343392</td>\n",
       "      <td>18.335718</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>520.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>rangeCity</th>\n",
       "      <td>44075.0</td>\n",
       "      <td>1.294496</td>\n",
       "      <td>18.369160</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>520.800000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>rangeCityA</th>\n",
       "      <td>44075.0</td>\n",
       "      <td>0.156062</td>\n",
       "      <td>2.674805</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>135.280000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>rangeHwy</th>\n",
       "      <td>44075.0</td>\n",
       "      <td>1.221439</td>\n",
       "      <td>17.598063</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>520.500000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>rangeHwyA</th>\n",
       "      <td>44075.0</td>\n",
       "      <td>0.145662</td>\n",
       "      <td>2.402991</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>114.760000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>UCity</th>\n",
       "      <td>44075.0</td>\n",
       "      <td>23.748747</td>\n",
       "      <td>12.848973</td>\n",
       "      <td>0.00</td>\n",
       "      <td>18.449150</td>\n",
       "      <td>21.7000</td>\n",
       "      <td>26.091300</td>\n",
       "      <td>224.800000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>UCityA</th>\n",
       "      <td>44075.0</td>\n",
       "      <td>1.077916</td>\n",
       "      <td>8.868052</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>207.262200</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>UHighway</th>\n",
       "      <td>44075.0</td>\n",
       "      <td>34.884815</td>\n",
       "      <td>12.273705</td>\n",
       "      <td>0.00</td>\n",
       "      <td>28.000000</td>\n",
       "      <td>33.4229</td>\n",
       "      <td>39.500000</td>\n",
       "      <td>187.100000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>UHighwayA</th>\n",
       "      <td>44075.0</td>\n",
       "      <td>1.310887</td>\n",
       "      <td>8.950533</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>173.143600</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>year</th>\n",
       "      <td>44075.0</td>\n",
       "      <td>2002.823914</td>\n",
       "      <td>11.798301</td>\n",
       "      <td>1984.00</td>\n",
       "      <td>1992.000000</td>\n",
       "      <td>2004.0000</td>\n",
       "      <td>2013.000000</td>\n",
       "      <td>2022.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>youSaveSpend</th>\n",
       "      <td>44075.0</td>\n",
       "      <td>-4416.863301</td>\n",
       "      <td>3706.878507</td>\n",
       "      <td>-32750.00</td>\n",
       "      <td>-6500.000000</td>\n",
       "      <td>-4500.0000</td>\n",
       "      <td>-2000.000000</td>\n",
       "      <td>6500.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>charge240b</th>\n",
       "      <td>44075.0</td>\n",
       "      <td>0.014512</td>\n",
       "      <td>0.306402</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>9.600000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>phevCity</th>\n",
       "      <td>44075.0</td>\n",
       "      <td>0.252275</td>\n",
       "      <td>3.667847</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>97.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>phevHwy</th>\n",
       "      <td>44075.0</td>\n",
       "      <td>0.254317</td>\n",
       "      <td>3.545343</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>81.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>phevComb</th>\n",
       "      <td>44075.0</td>\n",
       "      <td>0.252116</td>\n",
       "      <td>3.590077</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>88.000000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                   count          mean           std       min           25%  \\\n",
       "barrels08        44075.0     17.089905      4.698712      0.06     14.330870   \n",
       "barrelsA08       44075.0      0.222572      1.143200      0.00      0.000000   \n",
       "charge120        44075.0      0.000000      0.000000      0.00      0.000000   \n",
       "charge240        44075.0      0.073465      0.784051      0.00      0.000000   \n",
       "city08           44075.0     18.749223      9.036866      6.00     15.000000   \n",
       "city08U          44075.0      7.185707     12.972132      0.00      0.000000   \n",
       "cityA08          44075.0      0.814770      6.280906      0.00      0.000000   \n",
       "cityA08U         44075.0      0.679923      6.181746      0.00      0.000000   \n",
       "cityCD           44075.0      0.000501      0.037958      0.00      0.000000   \n",
       "cityE            44075.0      0.520147      4.940315      0.00      0.000000   \n",
       "cityUF           44075.0      0.002852      0.039264      0.00      0.000000   \n",
       "co2              44075.0    107.957686    187.773444     -1.00     -1.000000   \n",
       "co2A             44075.0      5.843607     57.032958     -1.00     -1.000000   \n",
       "co2TailpipeAGpm  44075.0     16.994214     91.955091      0.00      0.000000   \n",
       "co2TailpipeGpm   44075.0    461.017357    125.764745      0.00    386.391304   \n",
       "comb08           44075.0     20.985071      8.652291      7.00     17.000000   \n",
       "comb08U          44075.0      7.972696     13.548087      0.00      0.000000   \n",
       "combA08          44075.0      0.869541      6.250612      0.00      0.000000   \n",
       "combA08U         44075.0      0.713475      6.110493      0.00      0.000000   \n",
       "combE            44075.0      0.528292      4.962906      0.00      0.000000   \n",
       "combinedCD       44075.0      0.000389      0.033056      0.00      0.000000   \n",
       "combinedUF       44075.0      0.002825      0.038834      0.00      0.000000   \n",
       "cylinders        43770.0      5.708933      1.766317      2.00      4.000000   \n",
       "displ            43772.0      3.282066      1.356857      0.00      2.200000   \n",
       "engId            44075.0   7640.478775  16792.609574      0.00      0.000000   \n",
       "feScore          44075.0      0.656858      2.881343     -1.00     -1.000000   \n",
       "fuelCost08       44075.0   2636.197391    738.408030    450.00   2150.000000   \n",
       "fuelCostA08      44075.0     98.241634    525.379257      0.00      0.000000   \n",
       "ghgScore         44075.0      0.658015      2.887266     -1.00     -1.000000   \n",
       "ghgScoreA        44075.0     -0.922042      0.651828     -1.00     -1.000000   \n",
       "highway08        44075.0     24.853613      8.484246      9.00     20.000000   \n",
       "highway08U       44075.0      9.335541     14.939994      0.00      0.000000   \n",
       "highwayA08       44075.0      0.967487      6.381416      0.00      0.000000   \n",
       "highwayA08U      44075.0      0.779874      6.178836      0.00      0.000000   \n",
       "highwayCD        44075.0      0.000250      0.027495      0.00      0.000000   \n",
       "highwayE         44075.0      0.538645      5.012180      0.00      0.000000   \n",
       "highwayUF        44075.0      0.002788      0.038284      0.00      0.000000   \n",
       "hlv              44075.0      1.975201      5.912968      0.00      0.000000   \n",
       "hpv              44075.0     10.052706     27.812047      0.00      0.000000   \n",
       "id               44075.0  22179.946001  12869.366833      1.00  11019.500000   \n",
       "lv2              44075.0      1.769212      4.314686      0.00      0.000000   \n",
       "lv4              44075.0      6.034532      9.557379      0.00      0.000000   \n",
       "pv2              44075.0     13.315553     30.862476      0.00      0.000000   \n",
       "pv4              44075.0     33.536948     46.017789      0.00      0.000000   \n",
       "range            44075.0      1.343392     18.335718      0.00      0.000000   \n",
       "rangeCity        44075.0      1.294496     18.369160      0.00      0.000000   \n",
       "rangeCityA       44075.0      0.156062      2.674805      0.00      0.000000   \n",
       "rangeHwy         44075.0      1.221439     17.598063      0.00      0.000000   \n",
       "rangeHwyA        44075.0      0.145662      2.402991      0.00      0.000000   \n",
       "UCity            44075.0     23.748747     12.848973      0.00     18.449150   \n",
       "UCityA           44075.0      1.077916      8.868052      0.00      0.000000   \n",
       "UHighway         44075.0     34.884815     12.273705      0.00     28.000000   \n",
       "UHighwayA        44075.0      1.310887      8.950533      0.00      0.000000   \n",
       "year             44075.0   2002.823914     11.798301   1984.00   1992.000000   \n",
       "youSaveSpend     44075.0  -4416.863301   3706.878507 -32750.00  -6500.000000   \n",
       "charge240b       44075.0      0.014512      0.306402      0.00      0.000000   \n",
       "phevCity         44075.0      0.252275      3.667847      0.00      0.000000   \n",
       "phevHwy          44075.0      0.254317      3.545343      0.00      0.000000   \n",
       "phevComb         44075.0      0.252116      3.590077      0.00      0.000000   \n",
       "\n",
       "                        50%           75%           max  \n",
       "barrels08           16.4805     19.388824     47.087143  \n",
       "barrelsA08           0.0000      0.000000     18.311667  \n",
       "charge120            0.0000      0.000000      0.000000  \n",
       "charge240            0.0000      0.000000     15.300000  \n",
       "city08              17.0000     21.000000    150.000000  \n",
       "city08U              0.0000     16.169200    150.195800  \n",
       "cityA08              0.0000      0.000000    145.000000  \n",
       "cityA08U             0.0000      0.000000    145.083500  \n",
       "cityCD               0.0000      0.000000      5.350000  \n",
       "cityE                0.0000      0.000000    122.000000  \n",
       "cityUF               0.0000      0.000000      0.927000  \n",
       "co2                 -1.0000    275.000000    893.000000  \n",
       "co2A                -1.0000     -1.000000    713.000000  \n",
       "co2TailpipeAGpm      0.0000      0.000000    713.000000  \n",
       "co2TailpipeGpm     444.3500    522.764706   1269.571429  \n",
       "comb08              20.0000     23.000000    142.000000  \n",
       "comb08U              0.0000     18.632100    141.951200  \n",
       "combA08              0.0000      0.000000    133.000000  \n",
       "combA08U             0.0000      0.000000    133.266200  \n",
       "combE                0.0000      0.000000    121.000000  \n",
       "combinedCD           0.0000      0.000000      4.800000  \n",
       "combinedUF           0.0000      0.000000      0.920000  \n",
       "cylinders            6.0000      6.000000     16.000000  \n",
       "displ                3.0000      4.200000      8.400000  \n",
       "engId              162.0000   4117.000000  69102.000000  \n",
       "feScore             -1.0000      3.000000     10.000000  \n",
       "fuelCost08        2650.0000   3050.000000   8300.000000  \n",
       "fuelCostA08          0.0000      0.000000   4000.000000  \n",
       "ghgScore            -1.0000      3.000000     10.000000  \n",
       "ghgScoreA           -1.0000     -1.000000      8.000000  \n",
       "highway08           24.0000     28.000000    133.000000  \n",
       "highway08U           0.0000     22.780350    133.026300  \n",
       "highwayA08           0.0000      0.000000    121.000000  \n",
       "highwayA08U          0.0000      0.000000    121.200500  \n",
       "highwayCD            0.0000      0.000000      4.060000  \n",
       "highwayE             0.0000      0.000000    120.000000  \n",
       "highwayUF            0.0000      0.000000      0.910000  \n",
       "hlv                  0.0000      0.000000     49.000000  \n",
       "hpv                  0.0000      0.000000    195.000000  \n",
       "id               22042.0000  33359.500000  44497.000000  \n",
       "lv2                  0.0000      0.000000     41.000000  \n",
       "lv4                  0.0000     13.000000     55.000000  \n",
       "pv2                  0.0000      0.000000    194.000000  \n",
       "pv4                  0.0000     91.000000    192.000000  \n",
       "range                0.0000      0.000000    520.000000  \n",
       "rangeCity            0.0000      0.000000    520.800000  \n",
       "rangeCityA           0.0000      0.000000    135.280000  \n",
       "rangeHwy             0.0000      0.000000    520.500000  \n",
       "rangeHwyA            0.0000      0.000000    114.760000  \n",
       "UCity               21.7000     26.091300    224.800000  \n",
       "UCityA               0.0000      0.000000    207.262200  \n",
       "UHighway            33.4229     39.500000    187.100000  \n",
       "UHighwayA            0.0000      0.000000    173.143600  \n",
       "year              2004.0000   2013.000000   2022.000000  \n",
       "youSaveSpend     -4500.0000  -2000.000000   6500.000000  \n",
       "charge240b           0.0000      0.000000      9.600000  \n",
       "phevCity             0.0000      0.000000     97.000000  \n",
       "phevHwy              0.0000      0.000000     81.000000  \n",
       "phevComb             0.0000      0.000000     88.000000  "
      ]
     },
     "execution_count": 137,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Get vehicle data info and description\n",
    "vehicle_df.info()\n",
    "vehicle_df.describe().T"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "701b7129",
   "metadata": {},
   "source": [
    "We observe that there are several missing values (NaNs) in the vehicle data. However, there are even more NaN values because some features have -1 or 0 as values which represent no data. So, we have to replace these -1s and 0s with NaN.   "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 138,
   "id": "278e1a71",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "RangeIndex: 44075 entries, 0 to 44074\n",
      "Data columns (total 83 columns):\n",
      " #   Column           Non-Null Count  Dtype  \n",
      "---  ------           --------------  -----  \n",
      " 0   barrels08        44075 non-null  float64\n",
      " 1   barrelsA08       1764 non-null   float64\n",
      " 2   charge120        0 non-null      float64\n",
      " 3   charge240        532 non-null    float64\n",
      " 4   city08           44075 non-null  float64\n",
      " 5   city08U          14419 non-null  float64\n",
      " 6   cityA08          1764 non-null   float64\n",
      " 7   cityA08U         1214 non-null   float64\n",
      " 8   cityCD           45 non-null     float64\n",
      " 9   cityE            555 non-null    float64\n",
      " 10  cityUF           253 non-null    float64\n",
      " 11  co2              11834 non-null  float64\n",
      " 12  co2A             650 non-null    float64\n",
      " 13  co2TailpipeAGpm  1511 non-null   float64\n",
      " 14  co2TailpipeGpm   43773 non-null  float64\n",
      " 15  comb08           44075 non-null  float64\n",
      " 16  comb08U          14419 non-null  float64\n",
      " 17  combA08          1764 non-null   float64\n",
      " 18  combA08U         1214 non-null   float64\n",
      " 19  combE            555 non-null    float64\n",
      " 20  combinedCD       47 non-null     float64\n",
      " 21  combinedUF       253 non-null    float64\n",
      " 22  cylinders        43770 non-null  float64\n",
      " 23  displ            43771 non-null  float64\n",
      " 24  drive            42889 non-null  object \n",
      " 25  engId            31478 non-null  float64\n",
      " 26  eng_dscr         27463 non-null  object \n",
      " 27  feScore          12063 non-null  float64\n",
      " 28  fuelCost08       44075 non-null  float64\n",
      " 29  fuelCostA08      1572 non-null   float64\n",
      " 30  fuelType         44075 non-null  object \n",
      " 31  fuelType1        44075 non-null  object \n",
      " 32  ghgScore         12063 non-null  float64\n",
      " 33  ghgScoreA        669 non-null    float64\n",
      " 34  highway08        44075 non-null  float64\n",
      " 35  highway08U       14419 non-null  float64\n",
      " 36  highwayA08       1764 non-null   float64\n",
      " 37  highwayA08U      1214 non-null   float64\n",
      " 38  highwayCD        30 non-null     float64\n",
      " 39  highwayE         555 non-null    float64\n",
      " 40  highwayUF        253 non-null    float64\n",
      " 41  hlv              5148 non-null   float64\n",
      " 42  hpv              5147 non-null   float64\n",
      " 43  id               44075 non-null  float64\n",
      " 44  lv2              7083 non-null   float64\n",
      " 45  lv4              15416 non-null  float64\n",
      " 46  make             44075 non-null  object \n",
      " 47  model            44075 non-null  object \n",
      " 48  mpgData          44040 non-null  object \n",
      " 49  phevBlended      44075 non-null  bool   \n",
      " 50  pv2              7077 non-null   float64\n",
      " 51  pv4              15416 non-null  float64\n",
      " 52  range            302 non-null    float64\n",
      " 53  rangeCity        267 non-null    float64\n",
      " 54  rangeCityA       253 non-null    float64\n",
      " 55  rangeHwy         267 non-null    float64\n",
      " 56  rangeHwyA        253 non-null    float64\n",
      " 57  trany            44064 non-null  object \n",
      " 58  UCity            44050 non-null  float64\n",
      " 59  UCityA           1718 non-null   float64\n",
      " 60  UHighway         44050 non-null  float64\n",
      " 61  UHighwayA        1692 non-null   float64\n",
      " 62  VClass           44075 non-null  object \n",
      " 63  year             44075 non-null  float64\n",
      " 64  youSaveSpend     43355 non-null  float64\n",
      " 65  guzzler          2597 non-null   object \n",
      " 66  trans_dscr       15044 non-null  object \n",
      " 67  tCharger         8514 non-null   object \n",
      " 68  sCharger         952 non-null    object \n",
      " 69  atvType          4141 non-null   object \n",
      " 70  fuelType2        1764 non-null   object \n",
      " 71  rangeA           1759 non-null   object \n",
      " 72  evMotor          1363 non-null   object \n",
      " 73  mfrCode          13267 non-null  object \n",
      " 74  c240Dscr         118 non-null    object \n",
      " 75  charge240b       108 non-null    float64\n",
      " 76  c240bDscr        112 non-null    object \n",
      " 77  createdOn        44075 non-null  object \n",
      " 78  modifiedOn       44075 non-null  object \n",
      " 79  startStop        12386 non-null  object \n",
      " 80  phevCity         253 non-null    float64\n",
      " 81  phevHwy          253 non-null    float64\n",
      " 82  phevComb         253 non-null    float64\n",
      "dtypes: bool(1), float64(59), object(23)\n",
      "memory usage: 27.6+ MB\n"
     ]
    },
    {
     "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>count</th>\n",
       "      <th>mean</th>\n",
       "      <th>std</th>\n",
       "      <th>min</th>\n",
       "      <th>25%</th>\n",
       "      <th>50%</th>\n",
       "      <th>75%</th>\n",
       "      <th>max</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>barrels08</th>\n",
       "      <td>44075.0</td>\n",
       "      <td>17.089905</td>\n",
       "      <td>4.698712</td>\n",
       "      <td>0.06000</td>\n",
       "      <td>14.330870</td>\n",
       "      <td>16.480500</td>\n",
       "      <td>19.388824</td>\n",
       "      <td>47.087143</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>barrelsA08</th>\n",
       "      <td>1764.0</td>\n",
       "      <td>5.561138</td>\n",
       "      <td>1.722376</td>\n",
       "      <td>0.08087</td>\n",
       "      <td>4.681875</td>\n",
       "      <td>5.762308</td>\n",
       "      <td>6.810000</td>\n",
       "      <td>18.311667</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>charge120</th>\n",
       "      <td>0.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>charge240</th>\n",
       "      <td>532.0</td>\n",
       "      <td>6.086410</td>\n",
       "      <td>3.789202</td>\n",
       "      <td>0.67000</td>\n",
       "      <td>3.000000</td>\n",
       "      <td>4.000000</td>\n",
       "      <td>10.000000</td>\n",
       "      <td>15.300000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>city08</th>\n",
       "      <td>44075.0</td>\n",
       "      <td>18.749223</td>\n",
       "      <td>9.036866</td>\n",
       "      <td>6.00000</td>\n",
       "      <td>15.000000</td>\n",
       "      <td>17.000000</td>\n",
       "      <td>21.000000</td>\n",
       "      <td>150.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>city08U</th>\n",
       "      <td>14419.0</td>\n",
       "      <td>21.964772</td>\n",
       "      <td>13.775259</td>\n",
       "      <td>8.10000</td>\n",
       "      <td>16.329500</td>\n",
       "      <td>19.500000</td>\n",
       "      <td>23.302950</td>\n",
       "      <td>150.195800</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>cityA08</th>\n",
       "      <td>1764.0</td>\n",
       "      <td>20.357710</td>\n",
       "      <td>24.251744</td>\n",
       "      <td>7.00000</td>\n",
       "      <td>10.000000</td>\n",
       "      <td>12.000000</td>\n",
       "      <td>14.000000</td>\n",
       "      <td>145.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>cityA08U</th>\n",
       "      <td>1214.0</td>\n",
       "      <td>24.685014</td>\n",
       "      <td>28.203539</td>\n",
       "      <td>7.44300</td>\n",
       "      <td>10.611025</td>\n",
       "      <td>12.295300</td>\n",
       "      <td>16.035500</td>\n",
       "      <td>145.083500</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>cityCD</th>\n",
       "      <td>45.0</td>\n",
       "      <td>0.490809</td>\n",
       "      <td>1.094143</td>\n",
       "      <td>0.06030</td>\n",
       "      <td>0.100000</td>\n",
       "      <td>0.100000</td>\n",
       "      <td>0.500000</td>\n",
       "      <td>5.350000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>cityE</th>\n",
       "      <td>555.0</td>\n",
       "      <td>41.307192</td>\n",
       "      <td>15.932978</td>\n",
       "      <td>22.00000</td>\n",
       "      <td>28.264800</td>\n",
       "      <td>36.000000</td>\n",
       "      <td>50.183050</td>\n",
       "      <td>122.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>cityUF</th>\n",
       "      <td>253.0</td>\n",
       "      <td>0.496902</td>\n",
       "      <td>0.152203</td>\n",
       "      <td>0.24000</td>\n",
       "      <td>0.384000</td>\n",
       "      <td>0.453000</td>\n",
       "      <td>0.568000</td>\n",
       "      <td>0.927000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>co2</th>\n",
       "      <td>11834.0</td>\n",
       "      <td>404.780632</td>\n",
       "      <td>104.284015</td>\n",
       "      <td>22.00000</td>\n",
       "      <td>335.000000</td>\n",
       "      <td>397.000000</td>\n",
       "      <td>468.000000</td>\n",
       "      <td>893.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>co2A</th>\n",
       "      <td>650.0</td>\n",
       "      <td>463.049231</td>\n",
       "      <td>91.673572</td>\n",
       "      <td>270.00000</td>\n",
       "      <td>391.500000</td>\n",
       "      <td>457.000000</td>\n",
       "      <td>511.000000</td>\n",
       "      <td>713.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>co2TailpipeAGpm</th>\n",
       "      <td>1511.0</td>\n",
       "      <td>495.711435</td>\n",
       "      <td>96.664458</td>\n",
       "      <td>270.00000</td>\n",
       "      <td>419.666667</td>\n",
       "      <td>484.230769</td>\n",
       "      <td>572.272727</td>\n",
       "      <td>713.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>co2TailpipeGpm</th>\n",
       "      <td>43773.0</td>\n",
       "      <td>464.198021</td>\n",
       "      <td>120.205678</td>\n",
       "      <td>22.00000</td>\n",
       "      <td>386.391304</td>\n",
       "      <td>444.350000</td>\n",
       "      <td>522.764706</td>\n",
       "      <td>1269.571429</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>comb08</th>\n",
       "      <td>44075.0</td>\n",
       "      <td>20.985071</td>\n",
       "      <td>8.652291</td>\n",
       "      <td>7.00000</td>\n",
       "      <td>17.000000</td>\n",
       "      <td>20.000000</td>\n",
       "      <td>23.000000</td>\n",
       "      <td>142.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>comb08U</th>\n",
       "      <td>14419.0</td>\n",
       "      <td>24.370384</td>\n",
       "      <td>12.706117</td>\n",
       "      <td>9.95020</td>\n",
       "      <td>18.794850</td>\n",
       "      <td>22.161000</td>\n",
       "      <td>26.189550</td>\n",
       "      <td>141.951200</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>combA08</th>\n",
       "      <td>1764.0</td>\n",
       "      <td>21.726190</td>\n",
       "      <td>22.876605</td>\n",
       "      <td>9.00000</td>\n",
       "      <td>11.000000</td>\n",
       "      <td>13.000000</td>\n",
       "      <td>17.000000</td>\n",
       "      <td>133.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>combA08U</th>\n",
       "      <td>1214.0</td>\n",
       "      <td>25.903151</td>\n",
       "      <td>26.526211</td>\n",
       "      <td>8.64140</td>\n",
       "      <td>12.077100</td>\n",
       "      <td>14.100350</td>\n",
       "      <td>18.515000</td>\n",
       "      <td>133.266200</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>combE</th>\n",
       "      <td>555.0</td>\n",
       "      <td>41.953974</td>\n",
       "      <td>14.777980</td>\n",
       "      <td>23.74410</td>\n",
       "      <td>31.000000</td>\n",
       "      <td>36.230000</td>\n",
       "      <td>49.000000</td>\n",
       "      <td>121.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>combinedCD</th>\n",
       "      <td>47.0</td>\n",
       "      <td>0.365104</td>\n",
       "      <td>0.954392</td>\n",
       "      <td>0.09830</td>\n",
       "      <td>0.100000</td>\n",
       "      <td>0.100000</td>\n",
       "      <td>0.195350</td>\n",
       "      <td>4.800000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>combinedUF</th>\n",
       "      <td>253.0</td>\n",
       "      <td>0.492166</td>\n",
       "      <td>0.148193</td>\n",
       "      <td>0.23700</td>\n",
       "      <td>0.384000</td>\n",
       "      <td>0.450600</td>\n",
       "      <td>0.570000</td>\n",
       "      <td>0.920000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>cylinders</th>\n",
       "      <td>43770.0</td>\n",
       "      <td>5.708933</td>\n",
       "      <td>1.766317</td>\n",
       "      <td>2.00000</td>\n",
       "      <td>4.000000</td>\n",
       "      <td>6.000000</td>\n",
       "      <td>6.000000</td>\n",
       "      <td>16.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>displ</th>\n",
       "      <td>43771.0</td>\n",
       "      <td>3.282141</td>\n",
       "      <td>1.356782</td>\n",
       "      <td>0.60000</td>\n",
       "      <td>2.200000</td>\n",
       "      <td>3.000000</td>\n",
       "      <td>4.200000</td>\n",
       "      <td>8.400000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>engId</th>\n",
       "      <td>31478.0</td>\n",
       "      <td>10698.078086</td>\n",
       "      <td>19029.779272</td>\n",
       "      <td>1.00000</td>\n",
       "      <td>106.000000</td>\n",
       "      <td>1851.000000</td>\n",
       "      <td>4862.000000</td>\n",
       "      <td>69102.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>feScore</th>\n",
       "      <td>12063.0</td>\n",
       "      <td>5.053718</td>\n",
       "      <td>1.927710</td>\n",
       "      <td>1.00000</td>\n",
       "      <td>4.000000</td>\n",
       "      <td>5.000000</td>\n",
       "      <td>6.000000</td>\n",
       "      <td>10.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>fuelCost08</th>\n",
       "      <td>44075.0</td>\n",
       "      <td>2636.197391</td>\n",
       "      <td>738.408030</td>\n",
       "      <td>450.00000</td>\n",
       "      <td>2150.000000</td>\n",
       "      <td>2650.000000</td>\n",
       "      <td>3050.000000</td>\n",
       "      <td>8300.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>fuelCostA08</th>\n",
       "      <td>1572.0</td>\n",
       "      <td>2754.452926</td>\n",
       "      <td>650.152423</td>\n",
       "      <td>550.00000</td>\n",
       "      <td>2400.000000</td>\n",
       "      <td>2750.000000</td>\n",
       "      <td>3250.000000</td>\n",
       "      <td>4000.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>ghgScore</th>\n",
       "      <td>12063.0</td>\n",
       "      <td>5.057946</td>\n",
       "      <td>1.950313</td>\n",
       "      <td>1.00000</td>\n",
       "      <td>4.000000</td>\n",
       "      <td>5.000000</td>\n",
       "      <td>6.000000</td>\n",
       "      <td>10.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>ghgScoreA</th>\n",
       "      <td>669.0</td>\n",
       "      <td>4.118087</td>\n",
       "      <td>1.477208</td>\n",
       "      <td>1.00000</td>\n",
       "      <td>3.000000</td>\n",
       "      <td>4.000000</td>\n",
       "      <td>5.000000</td>\n",
       "      <td>8.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>highway08</th>\n",
       "      <td>44075.0</td>\n",
       "      <td>24.853613</td>\n",
       "      <td>8.484246</td>\n",
       "      <td>9.00000</td>\n",
       "      <td>20.000000</td>\n",
       "      <td>24.000000</td>\n",
       "      <td>28.000000</td>\n",
       "      <td>133.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>highway08U</th>\n",
       "      <td>14419.0</td>\n",
       "      <td>28.536234</td>\n",
       "      <td>11.590994</td>\n",
       "      <td>10.92320</td>\n",
       "      <td>22.965900</td>\n",
       "      <td>26.840700</td>\n",
       "      <td>31.048500</td>\n",
       "      <td>133.026300</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>highwayA08</th>\n",
       "      <td>1764.0</td>\n",
       "      <td>24.173469</td>\n",
       "      <td>21.371746</td>\n",
       "      <td>11.00000</td>\n",
       "      <td>13.000000</td>\n",
       "      <td>16.000000</td>\n",
       "      <td>21.000000</td>\n",
       "      <td>121.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>highwayA08U</th>\n",
       "      <td>1214.0</td>\n",
       "      <td>28.313802</td>\n",
       "      <td>24.636438</td>\n",
       "      <td>10.75870</td>\n",
       "      <td>14.910300</td>\n",
       "      <td>17.438800</td>\n",
       "      <td>22.541675</td>\n",
       "      <td>121.200500</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>highwayCD</th>\n",
       "      <td>30.0</td>\n",
       "      <td>0.366990</td>\n",
       "      <td>1.004849</td>\n",
       "      <td>0.00030</td>\n",
       "      <td>0.100000</td>\n",
       "      <td>0.100000</td>\n",
       "      <td>0.100000</td>\n",
       "      <td>4.060000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>highwayE</th>\n",
       "      <td>555.0</td>\n",
       "      <td>42.776145</td>\n",
       "      <td>13.732526</td>\n",
       "      <td>25.33710</td>\n",
       "      <td>33.000000</td>\n",
       "      <td>37.000000</td>\n",
       "      <td>50.000000</td>\n",
       "      <td>120.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>highwayUF</th>\n",
       "      <td>253.0</td>\n",
       "      <td>0.485621</td>\n",
       "      <td>0.144713</td>\n",
       "      <td>0.23400</td>\n",
       "      <td>0.380800</td>\n",
       "      <td>0.442000</td>\n",
       "      <td>0.563000</td>\n",
       "      <td>0.910000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>hlv</th>\n",
       "      <td>5148.0</td>\n",
       "      <td>16.910839</td>\n",
       "      <td>6.838730</td>\n",
       "      <td>4.00000</td>\n",
       "      <td>12.000000</td>\n",
       "      <td>16.000000</td>\n",
       "      <td>19.000000</td>\n",
       "      <td>49.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>hpv</th>\n",
       "      <td>5147.0</td>\n",
       "      <td>86.083738</td>\n",
       "      <td>8.864626</td>\n",
       "      <td>49.00000</td>\n",
       "      <td>81.000000</td>\n",
       "      <td>85.000000</td>\n",
       "      <td>92.000000</td>\n",
       "      <td>195.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>id</th>\n",
       "      <td>44075.0</td>\n",
       "      <td>22179.946001</td>\n",
       "      <td>12869.366833</td>\n",
       "      <td>1.00000</td>\n",
       "      <td>11019.500000</td>\n",
       "      <td>22042.000000</td>\n",
       "      <td>33359.500000</td>\n",
       "      <td>44497.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>lv2</th>\n",
       "      <td>7083.0</td>\n",
       "      <td>11.009177</td>\n",
       "      <td>3.757519</td>\n",
       "      <td>2.00000</td>\n",
       "      <td>9.000000</td>\n",
       "      <td>11.000000</td>\n",
       "      <td>13.000000</td>\n",
       "      <td>41.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>lv4</th>\n",
       "      <td>15416.0</td>\n",
       "      <td>17.252984</td>\n",
       "      <td>8.222022</td>\n",
       "      <td>1.00000</td>\n",
       "      <td>13.000000</td>\n",
       "      <td>15.000000</td>\n",
       "      <td>17.000000</td>\n",
       "      <td>55.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>pv2</th>\n",
       "      <td>7077.0</td>\n",
       "      <td>82.928218</td>\n",
       "      <td>12.612756</td>\n",
       "      <td>1.00000</td>\n",
       "      <td>78.000000</td>\n",
       "      <td>84.000000</td>\n",
       "      <td>90.000000</td>\n",
       "      <td>194.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>pv4</th>\n",
       "      <td>15416.0</td>\n",
       "      <td>95.883563</td>\n",
       "      <td>8.733047</td>\n",
       "      <td>4.00000</td>\n",
       "      <td>90.000000</td>\n",
       "      <td>96.000000</td>\n",
       "      <td>100.000000</td>\n",
       "      <td>192.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>range</th>\n",
       "      <td>302.0</td>\n",
       "      <td>196.059603</td>\n",
       "      <td>104.523423</td>\n",
       "      <td>29.00000</td>\n",
       "      <td>88.000000</td>\n",
       "      <td>215.000000</td>\n",
       "      <td>265.000000</td>\n",
       "      <td>520.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>rangeCity</th>\n",
       "      <td>267.0</td>\n",
       "      <td>213.688768</td>\n",
       "      <td>101.743844</td>\n",
       "      <td>42.40000</td>\n",
       "      <td>116.000000</td>\n",
       "      <td>233.700000</td>\n",
       "      <td>279.150000</td>\n",
       "      <td>520.800000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>rangeCityA</th>\n",
       "      <td>253.0</td>\n",
       "      <td>27.187424</td>\n",
       "      <td>22.660197</td>\n",
       "      <td>8.62000</td>\n",
       "      <td>15.350000</td>\n",
       "      <td>19.420000</td>\n",
       "      <td>27.760000</td>\n",
       "      <td>135.280000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>rangeHwy</th>\n",
       "      <td>267.0</td>\n",
       "      <td>201.628993</td>\n",
       "      <td>103.699039</td>\n",
       "      <td>31.62000</td>\n",
       "      <td>95.000000</td>\n",
       "      <td>217.400000</td>\n",
       "      <td>277.165000</td>\n",
       "      <td>520.500000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>rangeHwyA</th>\n",
       "      <td>253.0</td>\n",
       "      <td>25.375753</td>\n",
       "      <td>19.161040</td>\n",
       "      <td>8.37000</td>\n",
       "      <td>15.270000</td>\n",
       "      <td>19.049000</td>\n",
       "      <td>27.170000</td>\n",
       "      <td>114.760000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>UCity</th>\n",
       "      <td>44050.0</td>\n",
       "      <td>23.762225</td>\n",
       "      <td>12.840153</td>\n",
       "      <td>7.00000</td>\n",
       "      <td>18.477500</td>\n",
       "      <td>21.700000</td>\n",
       "      <td>26.093900</td>\n",
       "      <td>224.800000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>UCityA</th>\n",
       "      <td>1718.0</td>\n",
       "      <td>27.653754</td>\n",
       "      <td>35.823673</td>\n",
       "      <td>8.86160</td>\n",
       "      <td>12.086500</td>\n",
       "      <td>14.439700</td>\n",
       "      <td>17.666450</td>\n",
       "      <td>207.262200</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>UHighway</th>\n",
       "      <td>44050.0</td>\n",
       "      <td>34.904613</td>\n",
       "      <td>12.249011</td>\n",
       "      <td>12.00000</td>\n",
       "      <td>28.000000</td>\n",
       "      <td>33.455200</td>\n",
       "      <td>39.500000</td>\n",
       "      <td>187.100000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>UHighwayA</th>\n",
       "      <td>1692.0</td>\n",
       "      <td>34.147354</td>\n",
       "      <td>31.081959</td>\n",
       "      <td>13.45550</td>\n",
       "      <td>18.500000</td>\n",
       "      <td>22.949400</td>\n",
       "      <td>29.325000</td>\n",
       "      <td>173.143600</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>year</th>\n",
       "      <td>44075.0</td>\n",
       "      <td>2002.823914</td>\n",
       "      <td>11.798301</td>\n",
       "      <td>1984.00000</td>\n",
       "      <td>1992.000000</td>\n",
       "      <td>2004.000000</td>\n",
       "      <td>2013.000000</td>\n",
       "      <td>2022.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>youSaveSpend</th>\n",
       "      <td>43355.0</td>\n",
       "      <td>-4490.214508</td>\n",
       "      <td>3693.207322</td>\n",
       "      <td>-32750.00000</td>\n",
       "      <td>-6500.000000</td>\n",
       "      <td>-4500.000000</td>\n",
       "      <td>-2000.000000</td>\n",
       "      <td>6500.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>charge240b</th>\n",
       "      <td>108.0</td>\n",
       "      <td>5.922222</td>\n",
       "      <td>1.832168</td>\n",
       "      <td>2.30000</td>\n",
       "      <td>4.750000</td>\n",
       "      <td>4.750000</td>\n",
       "      <td>8.000000</td>\n",
       "      <td>9.600000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>phevCity</th>\n",
       "      <td>253.0</td>\n",
       "      <td>43.948617</td>\n",
       "      <td>20.612865</td>\n",
       "      <td>15.00000</td>\n",
       "      <td>28.000000</td>\n",
       "      <td>37.000000</td>\n",
       "      <td>57.000000</td>\n",
       "      <td>97.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>phevHwy</th>\n",
       "      <td>253.0</td>\n",
       "      <td>44.304348</td>\n",
       "      <td>15.459826</td>\n",
       "      <td>20.00000</td>\n",
       "      <td>32.000000</td>\n",
       "      <td>39.000000</td>\n",
       "      <td>54.000000</td>\n",
       "      <td>81.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>phevComb</th>\n",
       "      <td>253.0</td>\n",
       "      <td>43.920949</td>\n",
       "      <td>18.127242</td>\n",
       "      <td>17.00000</td>\n",
       "      <td>29.000000</td>\n",
       "      <td>38.000000</td>\n",
       "      <td>55.000000</td>\n",
       "      <td>88.000000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                   count          mean           std          min  \\\n",
       "barrels08        44075.0     17.089905      4.698712      0.06000   \n",
       "barrelsA08        1764.0      5.561138      1.722376      0.08087   \n",
       "charge120            0.0           NaN           NaN          NaN   \n",
       "charge240          532.0      6.086410      3.789202      0.67000   \n",
       "city08           44075.0     18.749223      9.036866      6.00000   \n",
       "city08U          14419.0     21.964772     13.775259      8.10000   \n",
       "cityA08           1764.0     20.357710     24.251744      7.00000   \n",
       "cityA08U          1214.0     24.685014     28.203539      7.44300   \n",
       "cityCD              45.0      0.490809      1.094143      0.06030   \n",
       "cityE              555.0     41.307192     15.932978     22.00000   \n",
       "cityUF             253.0      0.496902      0.152203      0.24000   \n",
       "co2              11834.0    404.780632    104.284015     22.00000   \n",
       "co2A               650.0    463.049231     91.673572    270.00000   \n",
       "co2TailpipeAGpm   1511.0    495.711435     96.664458    270.00000   \n",
       "co2TailpipeGpm   43773.0    464.198021    120.205678     22.00000   \n",
       "comb08           44075.0     20.985071      8.652291      7.00000   \n",
       "comb08U          14419.0     24.370384     12.706117      9.95020   \n",
       "combA08           1764.0     21.726190     22.876605      9.00000   \n",
       "combA08U          1214.0     25.903151     26.526211      8.64140   \n",
       "combE              555.0     41.953974     14.777980     23.74410   \n",
       "combinedCD          47.0      0.365104      0.954392      0.09830   \n",
       "combinedUF         253.0      0.492166      0.148193      0.23700   \n",
       "cylinders        43770.0      5.708933      1.766317      2.00000   \n",
       "displ            43771.0      3.282141      1.356782      0.60000   \n",
       "engId            31478.0  10698.078086  19029.779272      1.00000   \n",
       "feScore          12063.0      5.053718      1.927710      1.00000   \n",
       "fuelCost08       44075.0   2636.197391    738.408030    450.00000   \n",
       "fuelCostA08       1572.0   2754.452926    650.152423    550.00000   \n",
       "ghgScore         12063.0      5.057946      1.950313      1.00000   \n",
       "ghgScoreA          669.0      4.118087      1.477208      1.00000   \n",
       "highway08        44075.0     24.853613      8.484246      9.00000   \n",
       "highway08U       14419.0     28.536234     11.590994     10.92320   \n",
       "highwayA08        1764.0     24.173469     21.371746     11.00000   \n",
       "highwayA08U       1214.0     28.313802     24.636438     10.75870   \n",
       "highwayCD           30.0      0.366990      1.004849      0.00030   \n",
       "highwayE           555.0     42.776145     13.732526     25.33710   \n",
       "highwayUF          253.0      0.485621      0.144713      0.23400   \n",
       "hlv               5148.0     16.910839      6.838730      4.00000   \n",
       "hpv               5147.0     86.083738      8.864626     49.00000   \n",
       "id               44075.0  22179.946001  12869.366833      1.00000   \n",
       "lv2               7083.0     11.009177      3.757519      2.00000   \n",
       "lv4              15416.0     17.252984      8.222022      1.00000   \n",
       "pv2               7077.0     82.928218     12.612756      1.00000   \n",
       "pv4              15416.0     95.883563      8.733047      4.00000   \n",
       "range              302.0    196.059603    104.523423     29.00000   \n",
       "rangeCity          267.0    213.688768    101.743844     42.40000   \n",
       "rangeCityA         253.0     27.187424     22.660197      8.62000   \n",
       "rangeHwy           267.0    201.628993    103.699039     31.62000   \n",
       "rangeHwyA          253.0     25.375753     19.161040      8.37000   \n",
       "UCity            44050.0     23.762225     12.840153      7.00000   \n",
       "UCityA            1718.0     27.653754     35.823673      8.86160   \n",
       "UHighway         44050.0     34.904613     12.249011     12.00000   \n",
       "UHighwayA         1692.0     34.147354     31.081959     13.45550   \n",
       "year             44075.0   2002.823914     11.798301   1984.00000   \n",
       "youSaveSpend     43355.0  -4490.214508   3693.207322 -32750.00000   \n",
       "charge240b         108.0      5.922222      1.832168      2.30000   \n",
       "phevCity           253.0     43.948617     20.612865     15.00000   \n",
       "phevHwy            253.0     44.304348     15.459826     20.00000   \n",
       "phevComb           253.0     43.920949     18.127242     17.00000   \n",
       "\n",
       "                          25%           50%           75%           max  \n",
       "barrels08           14.330870     16.480500     19.388824     47.087143  \n",
       "barrelsA08           4.681875      5.762308      6.810000     18.311667  \n",
       "charge120                 NaN           NaN           NaN           NaN  \n",
       "charge240            3.000000      4.000000     10.000000     15.300000  \n",
       "city08              15.000000     17.000000     21.000000    150.000000  \n",
       "city08U             16.329500     19.500000     23.302950    150.195800  \n",
       "cityA08             10.000000     12.000000     14.000000    145.000000  \n",
       "cityA08U            10.611025     12.295300     16.035500    145.083500  \n",
       "cityCD               0.100000      0.100000      0.500000      5.350000  \n",
       "cityE               28.264800     36.000000     50.183050    122.000000  \n",
       "cityUF               0.384000      0.453000      0.568000      0.927000  \n",
       "co2                335.000000    397.000000    468.000000    893.000000  \n",
       "co2A               391.500000    457.000000    511.000000    713.000000  \n",
       "co2TailpipeAGpm    419.666667    484.230769    572.272727    713.000000  \n",
       "co2TailpipeGpm     386.391304    444.350000    522.764706   1269.571429  \n",
       "comb08              17.000000     20.000000     23.000000    142.000000  \n",
       "comb08U             18.794850     22.161000     26.189550    141.951200  \n",
       "combA08             11.000000     13.000000     17.000000    133.000000  \n",
       "combA08U            12.077100     14.100350     18.515000    133.266200  \n",
       "combE               31.000000     36.230000     49.000000    121.000000  \n",
       "combinedCD           0.100000      0.100000      0.195350      4.800000  \n",
       "combinedUF           0.384000      0.450600      0.570000      0.920000  \n",
       "cylinders            4.000000      6.000000      6.000000     16.000000  \n",
       "displ                2.200000      3.000000      4.200000      8.400000  \n",
       "engId              106.000000   1851.000000   4862.000000  69102.000000  \n",
       "feScore              4.000000      5.000000      6.000000     10.000000  \n",
       "fuelCost08        2150.000000   2650.000000   3050.000000   8300.000000  \n",
       "fuelCostA08       2400.000000   2750.000000   3250.000000   4000.000000  \n",
       "ghgScore             4.000000      5.000000      6.000000     10.000000  \n",
       "ghgScoreA            3.000000      4.000000      5.000000      8.000000  \n",
       "highway08           20.000000     24.000000     28.000000    133.000000  \n",
       "highway08U          22.965900     26.840700     31.048500    133.026300  \n",
       "highwayA08          13.000000     16.000000     21.000000    121.000000  \n",
       "highwayA08U         14.910300     17.438800     22.541675    121.200500  \n",
       "highwayCD            0.100000      0.100000      0.100000      4.060000  \n",
       "highwayE            33.000000     37.000000     50.000000    120.000000  \n",
       "highwayUF            0.380800      0.442000      0.563000      0.910000  \n",
       "hlv                 12.000000     16.000000     19.000000     49.000000  \n",
       "hpv                 81.000000     85.000000     92.000000    195.000000  \n",
       "id               11019.500000  22042.000000  33359.500000  44497.000000  \n",
       "lv2                  9.000000     11.000000     13.000000     41.000000  \n",
       "lv4                 13.000000     15.000000     17.000000     55.000000  \n",
       "pv2                 78.000000     84.000000     90.000000    194.000000  \n",
       "pv4                 90.000000     96.000000    100.000000    192.000000  \n",
       "range               88.000000    215.000000    265.000000    520.000000  \n",
       "rangeCity          116.000000    233.700000    279.150000    520.800000  \n",
       "rangeCityA          15.350000     19.420000     27.760000    135.280000  \n",
       "rangeHwy            95.000000    217.400000    277.165000    520.500000  \n",
       "rangeHwyA           15.270000     19.049000     27.170000    114.760000  \n",
       "UCity               18.477500     21.700000     26.093900    224.800000  \n",
       "UCityA              12.086500     14.439700     17.666450    207.262200  \n",
       "UHighway            28.000000     33.455200     39.500000    187.100000  \n",
       "UHighwayA           18.500000     22.949400     29.325000    173.143600  \n",
       "year              1992.000000   2004.000000   2013.000000   2022.000000  \n",
       "youSaveSpend     -6500.000000  -4500.000000  -2000.000000   6500.000000  \n",
       "charge240b           4.750000      4.750000      8.000000      9.600000  \n",
       "phevCity            28.000000     37.000000     57.000000     97.000000  \n",
       "phevHwy             32.000000     39.000000     54.000000     81.000000  \n",
       "phevComb            29.000000     38.000000     55.000000     88.000000  "
      ]
     },
     "execution_count": 138,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Replace -1 with NaNs\n",
    "vehicle_df = vehicle_df.replace([-1, 0], np.nan)\n",
    "vehicle_df.info()\n",
    "vehicle_df.describe().T"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 139,
   "id": "5e291fea",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "barrels08       0.000000\n",
       "barrelsA08     95.997731\n",
       "charge120     100.000000\n",
       "charge240      98.792967\n",
       "city08          0.000000\n",
       "                 ...    \n",
       "modifiedOn      0.000000\n",
       "startStop      71.897901\n",
       "phevCity       99.425978\n",
       "phevHwy        99.425978\n",
       "phevComb       99.425978\n",
       "Length: 83, dtype: float64"
      ]
     },
     "execution_count": 139,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Check NaN values for vehicle\n",
    "percent_nan = vehicle_df.isnull().sum() * 100 / vehicle_df.shape[0]\n",
    "percent_nan"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "6f12cc88",
   "metadata": {},
   "source": [
    "Since want to include details about all types of cars, we can use the above colums (some of which we identified to be of most use). We can simply drop some columns such as modifiedOn and createdOn as these will not be useful predictors."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 140,
   "id": "214ffa36",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "barrels08      0.0\n",
       "city08         0.0\n",
       "comb08         0.0\n",
       "fuelCost08     0.0\n",
       "fuelType       0.0\n",
       "fuelType1      0.0\n",
       "highway08      0.0\n",
       "id             0.0\n",
       "make           0.0\n",
       "model          0.0\n",
       "phevBlended    0.0\n",
       "VClass         0.0\n",
       "year           0.0\n",
       "createdOn      0.0\n",
       "modifiedOn     0.0\n",
       "dtype: float64"
      ]
     },
     "execution_count": 140,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Columns with no missing values\n",
    "non_nan_cols = percent_nan[percent_nan == 0]\n",
    "non_nan_cols"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 141,
   "id": "982329c7",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Index(['barrels08', 'city08', 'comb08', 'fuelCost08', 'fuelType', 'fuelType1',\n",
      "       'highway08', 'id', 'make', 'model', 'phevBlended', 'VClass', 'year'],\n",
      "      dtype='object')\n"
     ]
    },
    {
     "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>barrels08</th>\n",
       "      <th>city08</th>\n",
       "      <th>comb08</th>\n",
       "      <th>fuelCost08</th>\n",
       "      <th>fuelType</th>\n",
       "      <th>fuelType1</th>\n",
       "      <th>highway08</th>\n",
       "      <th>id</th>\n",
       "      <th>make</th>\n",
       "      <th>model</th>\n",
       "      <th>phevBlended</th>\n",
       "      <th>VClass</th>\n",
       "      <th>year</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>15.695714</td>\n",
       "      <td>19.0</td>\n",
       "      <td>21.0</td>\n",
       "      <td>2250.0</td>\n",
       "      <td>Regular</td>\n",
       "      <td>Regular Gasoline</td>\n",
       "      <td>25.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>Alfa Romeo</td>\n",
       "      <td>Spider Veloce 2000</td>\n",
       "      <td>False</td>\n",
       "      <td>Two Seaters</td>\n",
       "      <td>1985.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>29.964545</td>\n",
       "      <td>9.0</td>\n",
       "      <td>11.0</td>\n",
       "      <td>4300.0</td>\n",
       "      <td>Regular</td>\n",
       "      <td>Regular Gasoline</td>\n",
       "      <td>14.0</td>\n",
       "      <td>10.0</td>\n",
       "      <td>Ferrari</td>\n",
       "      <td>Testarossa</td>\n",
       "      <td>False</td>\n",
       "      <td>Two Seaters</td>\n",
       "      <td>1985.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>12.207778</td>\n",
       "      <td>23.0</td>\n",
       "      <td>27.0</td>\n",
       "      <td>1750.0</td>\n",
       "      <td>Regular</td>\n",
       "      <td>Regular Gasoline</td>\n",
       "      <td>33.0</td>\n",
       "      <td>100.0</td>\n",
       "      <td>Dodge</td>\n",
       "      <td>Charger</td>\n",
       "      <td>False</td>\n",
       "      <td>Subcompact Cars</td>\n",
       "      <td>1985.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>29.964545</td>\n",
       "      <td>10.0</td>\n",
       "      <td>11.0</td>\n",
       "      <td>4300.0</td>\n",
       "      <td>Regular</td>\n",
       "      <td>Regular Gasoline</td>\n",
       "      <td>12.0</td>\n",
       "      <td>1000.0</td>\n",
       "      <td>Dodge</td>\n",
       "      <td>B150/B250 Wagon 2WD</td>\n",
       "      <td>False</td>\n",
       "      <td>Vans</td>\n",
       "      <td>1985.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>17.347895</td>\n",
       "      <td>17.0</td>\n",
       "      <td>19.0</td>\n",
       "      <td>3050.0</td>\n",
       "      <td>Premium</td>\n",
       "      <td>Premium Gasoline</td>\n",
       "      <td>23.0</td>\n",
       "      <td>10000.0</td>\n",
       "      <td>Subaru</td>\n",
       "      <td>Legacy AWD Turbo</td>\n",
       "      <td>False</td>\n",
       "      <td>Compact Cars</td>\n",
       "      <td>1993.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   barrels08  city08  comb08  fuelCost08 fuelType         fuelType1  \\\n",
       "0  15.695714    19.0    21.0      2250.0  Regular  Regular Gasoline   \n",
       "1  29.964545     9.0    11.0      4300.0  Regular  Regular Gasoline   \n",
       "2  12.207778    23.0    27.0      1750.0  Regular  Regular Gasoline   \n",
       "3  29.964545    10.0    11.0      4300.0  Regular  Regular Gasoline   \n",
       "4  17.347895    17.0    19.0      3050.0  Premium  Premium Gasoline   \n",
       "\n",
       "   highway08       id        make                model  phevBlended  \\\n",
       "0       25.0      1.0  Alfa Romeo   Spider Veloce 2000        False   \n",
       "1       14.0     10.0     Ferrari           Testarossa        False   \n",
       "2       33.0    100.0       Dodge              Charger        False   \n",
       "3       12.0   1000.0       Dodge  B150/B250 Wagon 2WD        False   \n",
       "4       23.0  10000.0      Subaru     Legacy AWD Turbo        False   \n",
       "\n",
       "            VClass    year  \n",
       "0      Two Seaters  1985.0  \n",
       "1      Two Seaters  1985.0  \n",
       "2  Subcompact Cars  1985.0  \n",
       "3             Vans  1985.0  \n",
       "4     Compact Cars  1993.0  "
      ]
     },
     "execution_count": 141,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "selected_columns = non_nan_cols.index[:-2]\n",
    "print(selected_columns)\n",
    "vehicle_df = vehicle_df[selected_columns]\n",
    "vehicle_df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 142,
   "id": "b6fee8f7",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "RangeIndex: 42442 entries, 0 to 42441\n",
      "Data columns (total 8 columns):\n",
      " #   Column         Non-Null Count  Dtype  \n",
      "---  ------         --------------  -----  \n",
      " 0   efid           42442 non-null  object \n",
      " 1   id             42442 non-null  int64  \n",
      " 2   salesArea      42442 non-null  int64  \n",
      " 3   score          42442 non-null  float64\n",
      " 4   scoreAlt       42442 non-null  float64\n",
      " 5   smartwayScore  42442 non-null  int64  \n",
      " 6   standard       42442 non-null  object \n",
      " 7   stdText        42428 non-null  object \n",
      "dtypes: float64(2), int64(3), object(3)\n",
      "memory usage: 2.6+ MB\n"
     ]
    },
    {
     "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>count</th>\n",
       "      <th>mean</th>\n",
       "      <th>std</th>\n",
       "      <th>min</th>\n",
       "      <th>25%</th>\n",
       "      <th>50%</th>\n",
       "      <th>75%</th>\n",
       "      <th>max</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>id</th>\n",
       "      <td>42442.0</td>\n",
       "      <td>32373.193770</td>\n",
       "      <td>7237.823953</td>\n",
       "      <td>19332.0</td>\n",
       "      <td>25265.25</td>\n",
       "      <td>33315.0</td>\n",
       "      <td>38526.75</td>\n",
       "      <td>43921.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>salesArea</th>\n",
       "      <td>42442.0</td>\n",
       "      <td>5.057632</td>\n",
       "      <td>2.135504</td>\n",
       "      <td>1.0</td>\n",
       "      <td>3.00</td>\n",
       "      <td>7.0</td>\n",
       "      <td>7.00</td>\n",
       "      <td>9.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>score</th>\n",
       "      <td>42442.0</td>\n",
       "      <td>5.015079</td>\n",
       "      <td>1.876870</td>\n",
       "      <td>-12.0</td>\n",
       "      <td>5.00</td>\n",
       "      <td>5.0</td>\n",
       "      <td>6.00</td>\n",
       "      <td>10.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>scoreAlt</th>\n",
       "      <td>42442.0</td>\n",
       "      <td>-0.947576</td>\n",
       "      <td>0.550470</td>\n",
       "      <td>-1.0</td>\n",
       "      <td>-1.00</td>\n",
       "      <td>-1.0</td>\n",
       "      <td>-1.00</td>\n",
       "      <td>6.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>smartwayScore</th>\n",
       "      <td>42442.0</td>\n",
       "      <td>-0.570001</td>\n",
       "      <td>0.856694</td>\n",
       "      <td>-1.0</td>\n",
       "      <td>-1.00</td>\n",
       "      <td>-1.0</td>\n",
       "      <td>-1.00</td>\n",
       "      <td>2.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                 count          mean          std      min       25%      50%  \\\n",
       "id             42442.0  32373.193770  7237.823953  19332.0  25265.25  33315.0   \n",
       "salesArea      42442.0      5.057632     2.135504      1.0      3.00      7.0   \n",
       "score          42442.0      5.015079     1.876870    -12.0      5.00      5.0   \n",
       "scoreAlt       42442.0     -0.947576     0.550470     -1.0     -1.00     -1.0   \n",
       "smartwayScore  42442.0     -0.570001     0.856694     -1.0     -1.00     -1.0   \n",
       "\n",
       "                    75%      max  \n",
       "id             38526.75  43921.0  \n",
       "salesArea          7.00      9.0  \n",
       "score              6.00     10.0  \n",
       "scoreAlt          -1.00      6.0  \n",
       "smartwayScore     -1.00      2.0  "
      ]
     },
     "execution_count": 142,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Get emissions data info and description\n",
    "emissions_df.info()\n",
    "emissions_df.describe().T"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "0f7d1efa",
   "metadata": {},
   "source": [
    "Here, we only need the id and score columns. So, we can drop the rest."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 143,
   "id": "ef906b3b",
   "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>count</th>\n",
       "      <th>mean</th>\n",
       "      <th>std</th>\n",
       "      <th>min</th>\n",
       "      <th>25%</th>\n",
       "      <th>50%</th>\n",
       "      <th>75%</th>\n",
       "      <th>max</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>id</th>\n",
       "      <td>42442.0</td>\n",
       "      <td>32373.193770</td>\n",
       "      <td>7237.823953</td>\n",
       "      <td>19332.0</td>\n",
       "      <td>25265.25</td>\n",
       "      <td>33315.0</td>\n",
       "      <td>38526.75</td>\n",
       "      <td>43921.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>score</th>\n",
       "      <td>42442.0</td>\n",
       "      <td>5.015079</td>\n",
       "      <td>1.876870</td>\n",
       "      <td>-12.0</td>\n",
       "      <td>5.00</td>\n",
       "      <td>5.0</td>\n",
       "      <td>6.00</td>\n",
       "      <td>10.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "         count          mean          std      min       25%      50%  \\\n",
       "id     42442.0  32373.193770  7237.823953  19332.0  25265.25  33315.0   \n",
       "score  42442.0      5.015079     1.876870    -12.0      5.00      5.0   \n",
       "\n",
       "            75%      max  \n",
       "id     38526.75  43921.0  \n",
       "score      6.00     10.0  "
      ]
     },
     "execution_count": 143,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "emissions_df = emissions_df[['id', 'score']]\n",
    "emissions_df.describe().T"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "1889ada6",
   "metadata": {},
   "source": [
    "Although, we have no missing values, the score column has -12.0 as the minimum value implying there might be incorrect values lying outisde the [1, 10] interval."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 144,
   "id": "47d88bd1",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "There are 20 incorrect emission scores\n"
     ]
    }
   ],
   "source": [
    "incorrect_scores = emissions_df.score[emissions_df.score < 1].size\n",
    "print('There are {} incorrect emission scores'.format(incorrect_scores))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 145,
   "id": "f90b5401",
   "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>count</th>\n",
       "      <th>mean</th>\n",
       "      <th>std</th>\n",
       "      <th>min</th>\n",
       "      <th>25%</th>\n",
       "      <th>50%</th>\n",
       "      <th>75%</th>\n",
       "      <th>max</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>id</th>\n",
       "      <td>42422.0</td>\n",
       "      <td>32378.618995</td>\n",
       "      <td>7235.194428</td>\n",
       "      <td>19332.0</td>\n",
       "      <td>25273.25</td>\n",
       "      <td>33320.0</td>\n",
       "      <td>38529.0</td>\n",
       "      <td>43921.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>score</th>\n",
       "      <td>42422.0</td>\n",
       "      <td>5.021451</td>\n",
       "      <td>1.851329</td>\n",
       "      <td>1.0</td>\n",
       "      <td>5.00</td>\n",
       "      <td>5.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>10.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "         count          mean          std      min       25%      50%  \\\n",
       "id     42422.0  32378.618995  7235.194428  19332.0  25273.25  33320.0   \n",
       "score  42422.0      5.021451     1.851329      1.0      5.00      5.0   \n",
       "\n",
       "           75%      max  \n",
       "id     38529.0  43921.0  \n",
       "score      6.0     10.0  "
      ]
     },
     "execution_count": 145,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Remove incorrect scores\n",
    "emissions_df = emissions_df[~(emissions_df.score < 1)]\n",
    "emissions_df.describe().T"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "e28a058f",
   "metadata": {},
   "source": [
    "Now, we have to join the two data frames based on id."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 146,
   "id": "4bd3f0ab",
   "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>barrels08</th>\n",
       "      <th>city08</th>\n",
       "      <th>comb08</th>\n",
       "      <th>fuelCost08</th>\n",
       "      <th>fuelType</th>\n",
       "      <th>fuelType1</th>\n",
       "      <th>highway08</th>\n",
       "      <th>id</th>\n",
       "      <th>make</th>\n",
       "      <th>model</th>\n",
       "      <th>phevBlended</th>\n",
       "      <th>VClass</th>\n",
       "      <th>year</th>\n",
       "      <th>score</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>18.311667</td>\n",
       "      <td>16.0</td>\n",
       "      <td>18.0</td>\n",
       "      <td>3200.0</td>\n",
       "      <td>Premium</td>\n",
       "      <td>Premium Gasoline</td>\n",
       "      <td>22.0</td>\n",
       "      <td>19332.0</td>\n",
       "      <td>Acura</td>\n",
       "      <td>NSX</td>\n",
       "      <td>False</td>\n",
       "      <td>Two Seaters</td>\n",
       "      <td>2004.0</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>18.311667</td>\n",
       "      <td>16.0</td>\n",
       "      <td>18.0</td>\n",
       "      <td>3200.0</td>\n",
       "      <td>Premium</td>\n",
       "      <td>Premium Gasoline</td>\n",
       "      <td>22.0</td>\n",
       "      <td>19332.0</td>\n",
       "      <td>Acura</td>\n",
       "      <td>NSX</td>\n",
       "      <td>False</td>\n",
       "      <td>Two Seaters</td>\n",
       "      <td>2004.0</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>18.311667</td>\n",
       "      <td>16.0</td>\n",
       "      <td>18.0</td>\n",
       "      <td>3200.0</td>\n",
       "      <td>Premium</td>\n",
       "      <td>Premium Gasoline</td>\n",
       "      <td>22.0</td>\n",
       "      <td>19333.0</td>\n",
       "      <td>Acura</td>\n",
       "      <td>NSX</td>\n",
       "      <td>False</td>\n",
       "      <td>Two Seaters</td>\n",
       "      <td>2004.0</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>18.311667</td>\n",
       "      <td>16.0</td>\n",
       "      <td>18.0</td>\n",
       "      <td>3200.0</td>\n",
       "      <td>Premium</td>\n",
       "      <td>Premium Gasoline</td>\n",
       "      <td>22.0</td>\n",
       "      <td>19333.0</td>\n",
       "      <td>Acura</td>\n",
       "      <td>NSX</td>\n",
       "      <td>False</td>\n",
       "      <td>Two Seaters</td>\n",
       "      <td>2004.0</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>15.695714</td>\n",
       "      <td>18.0</td>\n",
       "      <td>21.0</td>\n",
       "      <td>2750.0</td>\n",
       "      <td>Premium</td>\n",
       "      <td>Premium Gasoline</td>\n",
       "      <td>26.0</td>\n",
       "      <td>19334.0</td>\n",
       "      <td>Audi</td>\n",
       "      <td>TT Roadster</td>\n",
       "      <td>False</td>\n",
       "      <td>Two Seaters</td>\n",
       "      <td>2004.0</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   barrels08  city08  comb08  fuelCost08 fuelType         fuelType1  \\\n",
       "0  18.311667    16.0    18.0      3200.0  Premium  Premium Gasoline   \n",
       "1  18.311667    16.0    18.0      3200.0  Premium  Premium Gasoline   \n",
       "2  18.311667    16.0    18.0      3200.0  Premium  Premium Gasoline   \n",
       "3  18.311667    16.0    18.0      3200.0  Premium  Premium Gasoline   \n",
       "4  15.695714    18.0    21.0      2750.0  Premium  Premium Gasoline   \n",
       "\n",
       "   highway08       id   make        model  phevBlended       VClass    year  \\\n",
       "0       22.0  19332.0  Acura          NSX        False  Two Seaters  2004.0   \n",
       "1       22.0  19332.0  Acura          NSX        False  Two Seaters  2004.0   \n",
       "2       22.0  19333.0  Acura          NSX        False  Two Seaters  2004.0   \n",
       "3       22.0  19333.0  Acura          NSX        False  Two Seaters  2004.0   \n",
       "4       26.0  19334.0   Audi  TT Roadster        False  Two Seaters  2004.0   \n",
       "\n",
       "   score  \n",
       "0    1.0  \n",
       "1    1.0  \n",
       "2    1.0  \n",
       "3    1.0  \n",
       "4    1.0  "
      ]
     },
     "execution_count": 146,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# We perform inner join because we only need matching car ids.\n",
    "vehicle_emissions_df = vehicle_df.join(emissions_df.set_index('id'), on='id', how='inner').reset_index(drop=True)\n",
    "vehicle_emissions_df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 147,
   "id": "50749b69",
   "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>barrels08</th>\n",
       "      <th>city08</th>\n",
       "      <th>comb08</th>\n",
       "      <th>fuelCost08</th>\n",
       "      <th>fuelType</th>\n",
       "      <th>fuelType1</th>\n",
       "      <th>highway08</th>\n",
       "      <th>make</th>\n",
       "      <th>model</th>\n",
       "      <th>phevBlended</th>\n",
       "      <th>VClass</th>\n",
       "      <th>year</th>\n",
       "      <th>score</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>18.311667</td>\n",
       "      <td>16.0</td>\n",
       "      <td>18.0</td>\n",
       "      <td>3200.0</td>\n",
       "      <td>Premium</td>\n",
       "      <td>Premium Gasoline</td>\n",
       "      <td>22.0</td>\n",
       "      <td>Acura</td>\n",
       "      <td>NSX</td>\n",
       "      <td>False</td>\n",
       "      <td>Two Seaters</td>\n",
       "      <td>2004.0</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>15.695714</td>\n",
       "      <td>18.0</td>\n",
       "      <td>21.0</td>\n",
       "      <td>2750.0</td>\n",
       "      <td>Premium</td>\n",
       "      <td>Premium Gasoline</td>\n",
       "      <td>26.0</td>\n",
       "      <td>Audi</td>\n",
       "      <td>TT Roadster</td>\n",
       "      <td>False</td>\n",
       "      <td>Two Seaters</td>\n",
       "      <td>2004.0</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>15.695714</td>\n",
       "      <td>18.0</td>\n",
       "      <td>21.0</td>\n",
       "      <td>2750.0</td>\n",
       "      <td>Premium</td>\n",
       "      <td>Premium Gasoline</td>\n",
       "      <td>26.0</td>\n",
       "      <td>Audi</td>\n",
       "      <td>TT Roadster quattro</td>\n",
       "      <td>False</td>\n",
       "      <td>Two Seaters</td>\n",
       "      <td>2004.0</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>15.695714</td>\n",
       "      <td>19.0</td>\n",
       "      <td>21.0</td>\n",
       "      <td>2750.0</td>\n",
       "      <td>Premium</td>\n",
       "      <td>Premium Gasoline</td>\n",
       "      <td>25.0</td>\n",
       "      <td>Audi</td>\n",
       "      <td>TT Roadster quattro</td>\n",
       "      <td>False</td>\n",
       "      <td>Two Seaters</td>\n",
       "      <td>2004.0</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>15.695714</td>\n",
       "      <td>18.0</td>\n",
       "      <td>21.0</td>\n",
       "      <td>2750.0</td>\n",
       "      <td>Premium</td>\n",
       "      <td>Premium Gasoline</td>\n",
       "      <td>26.0</td>\n",
       "      <td>BMW</td>\n",
       "      <td>Z4 Roadster</td>\n",
       "      <td>False</td>\n",
       "      <td>Two Seaters</td>\n",
       "      <td>2004.0</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   barrels08  city08  comb08  fuelCost08 fuelType         fuelType1  \\\n",
       "0  18.311667    16.0    18.0      3200.0  Premium  Premium Gasoline   \n",
       "1  15.695714    18.0    21.0      2750.0  Premium  Premium Gasoline   \n",
       "2  15.695714    18.0    21.0      2750.0  Premium  Premium Gasoline   \n",
       "3  15.695714    19.0    21.0      2750.0  Premium  Premium Gasoline   \n",
       "4  15.695714    18.0    21.0      2750.0  Premium  Premium Gasoline   \n",
       "\n",
       "   highway08   make                model  phevBlended       VClass    year  \\\n",
       "0       22.0  Acura                  NSX        False  Two Seaters  2004.0   \n",
       "1       26.0   Audi          TT Roadster        False  Two Seaters  2004.0   \n",
       "2       26.0   Audi  TT Roadster quattro        False  Two Seaters  2004.0   \n",
       "3       25.0   Audi  TT Roadster quattro        False  Two Seaters  2004.0   \n",
       "4       26.0    BMW          Z4 Roadster        False  Two Seaters  2004.0   \n",
       "\n",
       "   score  \n",
       "0    1.0  \n",
       "1    1.0  \n",
       "2    1.0  \n",
       "3    1.0  \n",
       "4    1.0  "
      ]
     },
     "execution_count": 147,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Remove ID column and drop duplicate rows\n",
    "vehicle_emissions_df = vehicle_emissions_df.drop(columns=['id'])\n",
    "vehicle_emissions_df = vehicle_emissions_df.drop_duplicates().reset_index(drop=True)\n",
    "vehicle_emissions_df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 148,
   "id": "a1e6d470",
   "metadata": {},
   "outputs": [],
   "source": [
    "vehicle_emissions_df.describe()\n",
    "vehicle_emissions_df.to_csv('data/Vehicle_Emissions.csv', index=False)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "bd6e9813",
   "metadata": {},
   "source": [
    "Now, we have a cleaned data set having 28556 non-redundant observations that combines both vehicles and emissions data sets.  "
   ]
  },
  {
   "cell_type": "markdown",
   "id": "41639db4",
   "metadata": {},
   "source": [
    "## 3.2 Data Transformation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 149,
   "id": "907c0091",
   "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>barrels08</th>\n",
       "      <th>city08</th>\n",
       "      <th>comb08</th>\n",
       "      <th>fuelCost08</th>\n",
       "      <th>fuelType</th>\n",
       "      <th>fuelType1</th>\n",
       "      <th>highway08</th>\n",
       "      <th>make</th>\n",
       "      <th>model</th>\n",
       "      <th>phevBlended</th>\n",
       "      <th>VClass</th>\n",
       "      <th>year</th>\n",
       "      <th>score</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>18.311667</td>\n",
       "      <td>16.0</td>\n",
       "      <td>18.0</td>\n",
       "      <td>3200.0</td>\n",
       "      <td>Premium</td>\n",
       "      <td>Premium Gasoline</td>\n",
       "      <td>22.0</td>\n",
       "      <td>Acura</td>\n",
       "      <td>NSX</td>\n",
       "      <td>False</td>\n",
       "      <td>Two Seaters</td>\n",
       "      <td>2004.0</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>15.695714</td>\n",
       "      <td>18.0</td>\n",
       "      <td>21.0</td>\n",
       "      <td>2750.0</td>\n",
       "      <td>Premium</td>\n",
       "      <td>Premium Gasoline</td>\n",
       "      <td>26.0</td>\n",
       "      <td>Audi</td>\n",
       "      <td>TT Roadster</td>\n",
       "      <td>False</td>\n",
       "      <td>Two Seaters</td>\n",
       "      <td>2004.0</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>15.695714</td>\n",
       "      <td>18.0</td>\n",
       "      <td>21.0</td>\n",
       "      <td>2750.0</td>\n",
       "      <td>Premium</td>\n",
       "      <td>Premium Gasoline</td>\n",
       "      <td>26.0</td>\n",
       "      <td>Audi</td>\n",
       "      <td>TT Roadster quattro</td>\n",
       "      <td>False</td>\n",
       "      <td>Two Seaters</td>\n",
       "      <td>2004.0</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>15.695714</td>\n",
       "      <td>19.0</td>\n",
       "      <td>21.0</td>\n",
       "      <td>2750.0</td>\n",
       "      <td>Premium</td>\n",
       "      <td>Premium Gasoline</td>\n",
       "      <td>25.0</td>\n",
       "      <td>Audi</td>\n",
       "      <td>TT Roadster quattro</td>\n",
       "      <td>False</td>\n",
       "      <td>Two Seaters</td>\n",
       "      <td>2004.0</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>15.695714</td>\n",
       "      <td>18.0</td>\n",
       "      <td>21.0</td>\n",
       "      <td>2750.0</td>\n",
       "      <td>Premium</td>\n",
       "      <td>Premium Gasoline</td>\n",
       "      <td>26.0</td>\n",
       "      <td>BMW</td>\n",
       "      <td>Z4 Roadster</td>\n",
       "      <td>False</td>\n",
       "      <td>Two Seaters</td>\n",
       "      <td>2004.0</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   barrels08  city08  comb08  fuelCost08 fuelType         fuelType1  \\\n",
       "0  18.311667    16.0    18.0      3200.0  Premium  Premium Gasoline   \n",
       "1  15.695714    18.0    21.0      2750.0  Premium  Premium Gasoline   \n",
       "2  15.695714    18.0    21.0      2750.0  Premium  Premium Gasoline   \n",
       "3  15.695714    19.0    21.0      2750.0  Premium  Premium Gasoline   \n",
       "4  15.695714    18.0    21.0      2750.0  Premium  Premium Gasoline   \n",
       "\n",
       "   highway08   make                model  phevBlended       VClass    year  \\\n",
       "0       22.0  Acura                  NSX        False  Two Seaters  2004.0   \n",
       "1       26.0   Audi          TT Roadster        False  Two Seaters  2004.0   \n",
       "2       26.0   Audi  TT Roadster quattro        False  Two Seaters  2004.0   \n",
       "3       25.0   Audi  TT Roadster quattro        False  Two Seaters  2004.0   \n",
       "4       26.0    BMW          Z4 Roadster        False  Two Seaters  2004.0   \n",
       "\n",
       "   score  \n",
       "0    1.0  \n",
       "1    1.0  \n",
       "2    1.0  \n",
       "3    1.0  \n",
       "4    1.0  "
      ]
     },
     "execution_count": 149,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Read the cleaned data to avoid recomputation\n",
    "ve_df = pd.read_csv('data/Vehicle_Emissions.csv')\n",
    "ve_df.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "03dbcb89",
   "metadata": {},
   "source": [
    "First we check VClass categorical variable. This is an imporant variable as vehicle size is directly responsible for mileage and hence, emission score. We have to aggregate this column in to a higher level because there are two many categories."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 150,
   "id": "8df7e999",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Compact Cars                          4239\n",
       "Midsize Cars                          4152\n",
       "Subcompact Cars                       2736\n",
       "Large Cars                            2227\n",
       "Sport Utility Vehicle - 4WD           2023\n",
       "Sport Utility Vehicle - 2WD           1618\n",
       "Two Seaters                           1602\n",
       "Small Sport Utility Vehicle 4WD       1496\n",
       "Small Station Wagons                  1172\n",
       "Standard Sport Utility Vehicle 4WD    1166\n",
       "Small Sport Utility Vehicle 2WD        996\n",
       "Minicompact Cars                       975\n",
       "Standard Pickup Trucks 2WD             914\n",
       "Standard Pickup Trucks 4WD             851\n",
       "Standard Sport Utility Vehicle 2WD     433\n",
       "Midsize Station Wagons                 365\n",
       "Small Pickup Trucks 2WD                351\n",
       "Minivan - 2WD                          324\n",
       "Small Pickup Trucks 4WD                273\n",
       "Vans, Cargo Type                       205\n",
       "Special Purpose Vehicle 2WD            189\n",
       "Vans, Passenger Type                   168\n",
       "Minivan - 4WD                           42\n",
       "Special Purpose Vehicle 4WD             39\n",
       "Name: VClass, dtype: int64"
      ]
     },
     "execution_count": 150,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "ve_df.VClass.value_counts()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 151,
   "id": "2a0be7e4",
   "metadata": {},
   "outputs": [],
   "source": [
    "# vclass_1 = [\n",
    "#     'Compact Cars',\n",
    "#     'Subcompact Cars',\n",
    "#     'Two Seaters',\n",
    "#     'Minicompact Cars'\n",
    "# ]\n",
    "# vclass_2 = [\n",
    "#     'Midsize Cars',\n",
    "#     'Utility Vehicle',\n",
    "# ]\n",
    "# for vclass in ve_df.VClass.unique():\n",
    "#     selection = ve_df['VClass'] == vclass\n",
    "#     if vclass in vclass_1:\n",
    "#         ve_df.loc[selection, 'VClass'] = 1\n",
    "#     elif any(v in vclass for v in vclass_2):\n",
    "#         ve_df.loc[selection, 'VClass'] = 2\n",
    "#     else:\n",
    "#         ve_df.loc[selection, 'VClass'] = 3\n",
    "# ve_df.VClass.value_counts()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 152,
   "id": "2c021bca",
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Other Vehicles            26418\n",
       "Standard Pickup Trucks     1765\n",
       "Vans                        373\n",
       "Name: VClass, dtype: int64"
      ]
     },
     "execution_count": 152,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "vclass_1 = [\n",
    "    'Standard Pickup Trucks 2WD',\n",
    "    'Standard Pickup Trucks 4WD'\n",
    "]\n",
    "vclass_2 = [\n",
    "    'Vans, Cargo Type',\n",
    "    'Vans, Passenger Type'\n",
    "]\n",
    "for vclass in ve_df.VClass.unique():\n",
    "    selection = ve_df['VClass'] == vclass\n",
    "    if vclass in vclass_1:\n",
    "        ve_df.loc[selection, 'VClass'] = 'Standard Pickup Trucks'\n",
    "    elif vclass in vclass_2:\n",
    "        ve_df.loc[selection, 'VClass'] = 'Vans'\n",
    "    else:\n",
    "        ve_df.loc[selection, 'VClass'] = 'Other Vehicles'\n",
    "\n",
    "ve_df.VClass.value_counts()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "219810c3",
   "metadata": {},
   "source": [
    "So, we have made 3 categories out of the initial 24! Now, we check the fuelType1 column"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 153,
   "id": "eeed81b7",
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Regular Gasoline     15596\n",
       "Premium Gasoline     12175\n",
       "Diesel                 365\n",
       "Electricity            254\n",
       "Midgrade Gasoline      136\n",
       "Natural Gas             30\n",
       "Name: fuelType1, dtype: int64"
      ]
     },
     "execution_count": 153,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "ve_df.fuelType1.value_counts()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "5b82dfa1",
   "metadata": {},
   "source": [
    "We can combine midgrade and regular gasolines as 'Regular Gasoline'. We can also combine 'Electricity' and 'Natural Gas' as 'Green Fuel'."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 154,
   "id": "41896c70",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Regular Gasoline    15732\n",
       "Premium Gasoline    12175\n",
       "Diesel                365\n",
       "Green Fuel            284\n",
       "Name: fuelType1, dtype: int64"
      ]
     },
     "execution_count": 154,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "for ft in ve_df.fuelType1.unique():\n",
    "    selection = ve_df['fuelType1'] == ft\n",
    "    if ft == 'Midgrade Gasoline':\n",
    "        ve_df.loc[selection, 'fuelType1'] = 'Regular Gasoline'\n",
    "    elif ft in ['Electricity', 'Natural Gas']:\n",
    "        ve_df.loc[selection, 'fuelType1'] = 'Green Fuel'\n",
    "ve_df.fuelType1.value_counts()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 155,
   "id": "2774f1ab",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "False    28344\n",
       "True       212\n",
       "Name: phevBlended, dtype: int64"
      ]
     },
     "execution_count": 155,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "ve_df.phevBlended.value_counts()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 156,
   "id": "5e9bc9d5",
   "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>barrels08</th>\n",
       "      <th>city08</th>\n",
       "      <th>comb08</th>\n",
       "      <th>fuelCost08</th>\n",
       "      <th>fuelType1</th>\n",
       "      <th>highway08</th>\n",
       "      <th>phevBlended</th>\n",
       "      <th>VClass</th>\n",
       "      <th>year</th>\n",
       "      <th>score</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>18.311667</td>\n",
       "      <td>16.0</td>\n",
       "      <td>18.0</td>\n",
       "      <td>3200.0</td>\n",
       "      <td>Premium Gasoline</td>\n",
       "      <td>22.0</td>\n",
       "      <td>False</td>\n",
       "      <td>Other Vehicles</td>\n",
       "      <td>2004</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>15.695714</td>\n",
       "      <td>18.0</td>\n",
       "      <td>21.0</td>\n",
       "      <td>2750.0</td>\n",
       "      <td>Premium Gasoline</td>\n",
       "      <td>26.0</td>\n",
       "      <td>False</td>\n",
       "      <td>Other Vehicles</td>\n",
       "      <td>2004</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>15.695714</td>\n",
       "      <td>18.0</td>\n",
       "      <td>21.0</td>\n",
       "      <td>2750.0</td>\n",
       "      <td>Premium Gasoline</td>\n",
       "      <td>26.0</td>\n",
       "      <td>False</td>\n",
       "      <td>Other Vehicles</td>\n",
       "      <td>2004</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>15.695714</td>\n",
       "      <td>19.0</td>\n",
       "      <td>21.0</td>\n",
       "      <td>2750.0</td>\n",
       "      <td>Premium Gasoline</td>\n",
       "      <td>25.0</td>\n",
       "      <td>False</td>\n",
       "      <td>Other Vehicles</td>\n",
       "      <td>2004</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>15.695714</td>\n",
       "      <td>18.0</td>\n",
       "      <td>21.0</td>\n",
       "      <td>2750.0</td>\n",
       "      <td>Premium Gasoline</td>\n",
       "      <td>26.0</td>\n",
       "      <td>False</td>\n",
       "      <td>Other Vehicles</td>\n",
       "      <td>2004</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   barrels08  city08  comb08  fuelCost08         fuelType1  highway08  \\\n",
       "0  18.311667    16.0    18.0      3200.0  Premium Gasoline       22.0   \n",
       "1  15.695714    18.0    21.0      2750.0  Premium Gasoline       26.0   \n",
       "2  15.695714    18.0    21.0      2750.0  Premium Gasoline       26.0   \n",
       "3  15.695714    19.0    21.0      2750.0  Premium Gasoline       25.0   \n",
       "4  15.695714    18.0    21.0      2750.0  Premium Gasoline       26.0   \n",
       "\n",
       "   phevBlended          VClass  year  score  \n",
       "0        False  Other Vehicles  2004      1  \n",
       "1        False  Other Vehicles  2004      1  \n",
       "2        False  Other Vehicles  2004      1  \n",
       "3        False  Other Vehicles  2004      1  \n",
       "4        False  Other Vehicles  2004      1  "
      ]
     },
     "execution_count": 156,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# We drop columns that won't be useful or are redundant\n",
    "ve_df = ve_df.drop(columns=['fuelType', 'make', 'model'])\n",
    "\n",
    "# Set int types to year and score\n",
    "ve_df.year = ve_df.year.astype(int)\n",
    "ve_df.score = ve_df.score.astype(int)\n",
    "\n",
    "ve_df.to_csv('data/VE_Transformed.csv', index=False)\n",
    "ve_df.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "548776d8",
   "metadata": {},
   "source": [
    "Note that, so far, we have handled missing values and categorical variables. We also handled incorrect values. Next, as a part of data visualization, we will handle outliers."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "080ce5c5",
   "metadata": {},
   "source": [
    "## 3.3 Dimensionality Reduction"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 157,
   "id": "59e53259",
   "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>barrels08</th>\n",
       "      <th>city08</th>\n",
       "      <th>comb08</th>\n",
       "      <th>fuelCost08</th>\n",
       "      <th>fuelType1</th>\n",
       "      <th>highway08</th>\n",
       "      <th>phevBlended</th>\n",
       "      <th>VClass</th>\n",
       "      <th>year</th>\n",
       "      <th>score</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>18.311667</td>\n",
       "      <td>16.0</td>\n",
       "      <td>18.0</td>\n",
       "      <td>3200.0</td>\n",
       "      <td>Premium Gasoline</td>\n",
       "      <td>22.0</td>\n",
       "      <td>False</td>\n",
       "      <td>Other Vehicles</td>\n",
       "      <td>2004</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>15.695714</td>\n",
       "      <td>18.0</td>\n",
       "      <td>21.0</td>\n",
       "      <td>2750.0</td>\n",
       "      <td>Premium Gasoline</td>\n",
       "      <td>26.0</td>\n",
       "      <td>False</td>\n",
       "      <td>Other Vehicles</td>\n",
       "      <td>2004</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>15.695714</td>\n",
       "      <td>18.0</td>\n",
       "      <td>21.0</td>\n",
       "      <td>2750.0</td>\n",
       "      <td>Premium Gasoline</td>\n",
       "      <td>26.0</td>\n",
       "      <td>False</td>\n",
       "      <td>Other Vehicles</td>\n",
       "      <td>2004</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>15.695714</td>\n",
       "      <td>19.0</td>\n",
       "      <td>21.0</td>\n",
       "      <td>2750.0</td>\n",
       "      <td>Premium Gasoline</td>\n",
       "      <td>25.0</td>\n",
       "      <td>False</td>\n",
       "      <td>Other Vehicles</td>\n",
       "      <td>2004</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>15.695714</td>\n",
       "      <td>18.0</td>\n",
       "      <td>21.0</td>\n",
       "      <td>2750.0</td>\n",
       "      <td>Premium Gasoline</td>\n",
       "      <td>26.0</td>\n",
       "      <td>False</td>\n",
       "      <td>Other Vehicles</td>\n",
       "      <td>2004</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   barrels08  city08  comb08  fuelCost08         fuelType1  highway08  \\\n",
       "0  18.311667    16.0    18.0      3200.0  Premium Gasoline       22.0   \n",
       "1  15.695714    18.0    21.0      2750.0  Premium Gasoline       26.0   \n",
       "2  15.695714    18.0    21.0      2750.0  Premium Gasoline       26.0   \n",
       "3  15.695714    19.0    21.0      2750.0  Premium Gasoline       25.0   \n",
       "4  15.695714    18.0    21.0      2750.0  Premium Gasoline       26.0   \n",
       "\n",
       "   phevBlended          VClass  year  score  \n",
       "0        False  Other Vehicles  2004      1  \n",
       "1        False  Other Vehicles  2004      1  \n",
       "2        False  Other Vehicles  2004      1  \n",
       "3        False  Other Vehicles  2004      1  \n",
       "4        False  Other Vehicles  2004      1  "
      ]
     },
     "execution_count": 157,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "ve_df = pd.read_csv('data/VE_Transformed.csv')\n",
    "ve_df.VClass = ve_df.VClass.astype(str)\n",
    "ve_df.phevBlended = ve_df.phevBlended.astype(bool)\n",
    "ve_df.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "abaa76e9",
   "metadata": {},
   "source": [
    "### 3.3.1 Data Tabularization"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 158,
   "id": "cd3ea8a1",
   "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>col_0</th>\n",
       "      <th>Percent</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>score</th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>6.821684</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>4.135733</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>6.604566</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>0.021011</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>39.900546</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>29.254798</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>6.149321</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>1.908531</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>4.321334</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>0.882477</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "col_0    Percent\n",
       "score           \n",
       "1       6.821684\n",
       "2       4.135733\n",
       "3       6.604566\n",
       "4       0.021011\n",
       "5      39.900546\n",
       "6      29.254798\n",
       "7       6.149321\n",
       "8       1.908531\n",
       "9       4.321334\n",
       "10      0.882477"
      ]
     },
     "execution_count": 158,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Relative frequency table\n",
    "\n",
    "pd.crosstab(index=ve_df.score, columns=\"Percent\") * 100 / pd.crosstab(index=ve_df.score, columns=\"Percent\").sum()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "12bc40fd",
   "metadata": {},
   "source": [
    "From the relative frequency table, we see that emission scores 5 and 6 are the most frequent."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 159,
   "id": "84683d1b",
   "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>score</th>\n",
       "      <th>1</th>\n",
       "      <th>2</th>\n",
       "      <th>3</th>\n",
       "      <th>4</th>\n",
       "      <th>5</th>\n",
       "      <th>6</th>\n",
       "      <th>7</th>\n",
       "      <th>8</th>\n",
       "      <th>9</th>\n",
       "      <th>10</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>fuelType1</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>Diesel</th>\n",
       "      <td>104</td>\n",
       "      <td>1</td>\n",
       "      <td>117</td>\n",
       "      <td>6</td>\n",
       "      <td>80</td>\n",
       "      <td>57</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Green Fuel</th>\n",
       "      <td>2</td>\n",
       "      <td>5</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>11</td>\n",
       "      <td>12</td>\n",
       "      <td>251</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Premium Gasoline</th>\n",
       "      <td>861</td>\n",
       "      <td>335</td>\n",
       "      <td>1139</td>\n",
       "      <td>0</td>\n",
       "      <td>5350</td>\n",
       "      <td>3333</td>\n",
       "      <td>657</td>\n",
       "      <td>206</td>\n",
       "      <td>293</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Regular Gasoline</th>\n",
       "      <td>981</td>\n",
       "      <td>840</td>\n",
       "      <td>630</td>\n",
       "      <td>0</td>\n",
       "      <td>5962</td>\n",
       "      <td>4963</td>\n",
       "      <td>1099</td>\n",
       "      <td>328</td>\n",
       "      <td>929</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "score              1    2     3   4     5     6     7    8    9    10\n",
       "fuelType1                                                            \n",
       "Diesel            104    1   117   6    80    57     0    0    0    0\n",
       "Green Fuel          2    5     0   0     2     1     0   11   12  251\n",
       "Premium Gasoline  861  335  1139   0  5350  3333   657  206  293    1\n",
       "Regular Gasoline  981  840   630   0  5962  4963  1099  328  929    0"
      ]
     },
     "execution_count": 159,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Contingency Table\n",
    "\n",
    "pd.crosstab(ve_df.fuelType1, ve_df.score, margins=False)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "08c3e7ce",
   "metadata": {},
   "source": [
    "The contingency table shows that Green Fuel vehicles mostly receive a score of 10 (cleanest). Premium Gasoline and Regular Gasoline vehicles have a wide range of scores. There are no diesel vehicles having an emission score of more than 6.  "
   ]
  },
  {
   "cell_type": "markdown",
   "id": "eb4d3749",
   "metadata": {},
   "source": [
    "### 3.3.2 Correlation Analysis"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 160,
   "id": "36e4cfbd",
   "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>barrels08</th>\n",
       "      <th>city08</th>\n",
       "      <th>comb08</th>\n",
       "      <th>fuelCost08</th>\n",
       "      <th>highway08</th>\n",
       "      <th>phevBlended</th>\n",
       "      <th>year</th>\n",
       "      <th>score</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>barrels08</th>\n",
       "      <td>1.000000</td>\n",
       "      <td>-0.736932</td>\n",
       "      <td>-0.780911</td>\n",
       "      <td>0.910707</td>\n",
       "      <td>-0.818702</td>\n",
       "      <td>-0.190832</td>\n",
       "      <td>-0.294785</td>\n",
       "      <td>-0.425487</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>city08</th>\n",
       "      <td>-0.736932</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.993217</td>\n",
       "      <td>-0.648146</td>\n",
       "      <td>0.955490</td>\n",
       "      <td>0.090112</td>\n",
       "      <td>0.225552</td>\n",
       "      <td>0.383834</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>comb08</th>\n",
       "      <td>-0.780911</td>\n",
       "      <td>0.993217</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>-0.690899</td>\n",
       "      <td>0.980921</td>\n",
       "      <td>0.079363</td>\n",
       "      <td>0.240110</td>\n",
       "      <td>0.396941</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>fuelCost08</th>\n",
       "      <td>0.910707</td>\n",
       "      <td>-0.648146</td>\n",
       "      <td>-0.690899</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>-0.726165</td>\n",
       "      <td>-0.075753</td>\n",
       "      <td>-0.216122</td>\n",
       "      <td>-0.385754</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>highway08</th>\n",
       "      <td>-0.818702</td>\n",
       "      <td>0.955490</td>\n",
       "      <td>0.980921</td>\n",
       "      <td>-0.726165</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.055124</td>\n",
       "      <td>0.252753</td>\n",
       "      <td>0.408956</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>phevBlended</th>\n",
       "      <td>-0.190832</td>\n",
       "      <td>0.090112</td>\n",
       "      <td>0.079363</td>\n",
       "      <td>-0.075753</td>\n",
       "      <td>0.055124</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.099232</td>\n",
       "      <td>0.064428</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>year</th>\n",
       "      <td>-0.294785</td>\n",
       "      <td>0.225552</td>\n",
       "      <td>0.240110</td>\n",
       "      <td>-0.216122</td>\n",
       "      <td>0.252753</td>\n",
       "      <td>0.099232</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.231791</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>score</th>\n",
       "      <td>-0.425487</td>\n",
       "      <td>0.383834</td>\n",
       "      <td>0.396941</td>\n",
       "      <td>-0.385754</td>\n",
       "      <td>0.408956</td>\n",
       "      <td>0.064428</td>\n",
       "      <td>0.231791</td>\n",
       "      <td>1.000000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "             barrels08    city08    comb08  fuelCost08  highway08  \\\n",
       "barrels08     1.000000 -0.736932 -0.780911    0.910707  -0.818702   \n",
       "city08       -0.736932  1.000000  0.993217   -0.648146   0.955490   \n",
       "comb08       -0.780911  0.993217  1.000000   -0.690899   0.980921   \n",
       "fuelCost08    0.910707 -0.648146 -0.690899    1.000000  -0.726165   \n",
       "highway08    -0.818702  0.955490  0.980921   -0.726165   1.000000   \n",
       "phevBlended  -0.190832  0.090112  0.079363   -0.075753   0.055124   \n",
       "year         -0.294785  0.225552  0.240110   -0.216122   0.252753   \n",
       "score        -0.425487  0.383834  0.396941   -0.385754   0.408956   \n",
       "\n",
       "             phevBlended      year     score  \n",
       "barrels08      -0.190832 -0.294785 -0.425487  \n",
       "city08          0.090112  0.225552  0.383834  \n",
       "comb08          0.079363  0.240110  0.396941  \n",
       "fuelCost08     -0.075753 -0.216122 -0.385754  \n",
       "highway08       0.055124  0.252753  0.408956  \n",
       "phevBlended     1.000000  0.099232  0.064428  \n",
       "year            0.099232  1.000000  0.231791  \n",
       "score           0.064428  0.231791  1.000000  "
      ]
     },
     "execution_count": 160,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Correlation table\n",
    "ve_df.corr(method='pearson')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 161,
   "id": "43439b55",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Correlation heatmap\n",
    "%matplotlib inline\n",
    "ve_corr = ve_df.corr()\n",
    "mask = np.zeros_like(ve_corr)\n",
    "mask[np.triu_indices_from(mask)] = True\n",
    "sns.heatmap(ve_corr, cmap='Spectral', vmax=1.0, vmin=-1.0 , mask = mask, linewidths=2)\n",
    "\n",
    "plt.yticks(rotation=0)\n",
    "plt.xticks(rotation=45)\n",
    "\n",
    "plt.savefig('correlation_heatmap.png', dpi=600, bbox_inches='tight')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a117bbe0",
   "metadata": {},
   "source": [
    "From the correlation analysis, we see that barrels08 (annual petroleum consumption in barrels) shows the strongest correlation with emisisons score, whereas, phevBlended (hybrid car or not) shows the least correlation.Intriguingly, we observe that most of our selected variables exhibit a relatively high correlation (either positive or negative) with the emissions score.  "
   ]
  },
  {
   "cell_type": "markdown",
   "id": "b108eb10",
   "metadata": {},
   "source": [
    "### 3.3.3 Distribution and Cross-Comparison Analysis"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 162,
   "id": "d90db279",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYIAAAEGCAYAAABo25JHAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjMuMiwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy8vihELAAAACXBIWXMAAAsTAAALEwEAmpwYAAA5R0lEQVR4nO3dd3ic1ZX48e+RRr1LltyNbGNjDLYpwgGTUEIIpoUkpJhsEshmFxxC2qaRZH/Zkt1NL2RDIISwm7KBJJSEYnpCNcUF417kLltW723a+f0xM0KWR9LM6H1VZs7neXjQvPPOvHdAes/cc+89V1QVY4wxqSttvBtgjDFmfFkgMMaYFGeBwBhjUpwFAmOMSXEWCIwxJsV5xrsB8ZoyZYpWVlaOdzOMMWZS2bBhQ6Oqlkd7ztVAICIrgduAdOBuVf3OoOe/DPzdgLacCpSravNQ71lZWcn69etdarExxiQnETk41HOupYZEJB24HbgcWAxcJyKLB56jqt9X1TNU9Qzga8DzwwUBY4wxznNzjGA5UK2q+1TVC9wHXDPM+dcB97rYHmOMMVG4GQhmAocHPK4JHzuBiOQCK4EHhnj+RhFZLyLrGxoaHG+oMcakMjcDgUQ5NlQ9i6uBl4dKC6nqXapapapV5eVRxzqMMcYkyM1AUAPMHvB4FnB0iHNXYWkhY4wZF24GgnXAAhGZKyKZhG72Dw8+SUSKgAuBv7jYFmOMMUNwbfqoqvpF5BbgSULTR+9R1W0isjr8/J3hU98HPKWqXW61xRhjzNBkspWhrqqqUltHYIwx8RGRDapaFe05KzFhzAD/789b+cT/vD7ezTBmTE26EhPGuKXb6+f+DTX0+gO0dnspzs0c7yYZMyasR2BM2F931tPjC6AKa/c2jXdzjBkzFgiMCXv0zVrKC7LIz/LwUnXjeDfHmDFjqSFjgI5eH3/dVc9Hls+hpqWHl/ZYIDCpw3oExgDbjrbj9Qe5eFEF71gwhUPN3Rxq6h7vZhkzJqxHYAxwrK0XgFklOUwrzAZg46EW5pTljmezjBkTFgiMAWrDgSASBACOtfeOV3OMGVMWCIwBjrX1UJjtIS8r9CdRkOWhzgKBSRE2RmAMoR7B9KKc/scVhVnUt/eNY4uMGTsWCIwhlAaaVvRWWmhqYbalhkzKsEBgDJEewfGBwFJDJlVYIDApz+sP0tjZd1yPIJIammxFGY1JhAUCk/LqO3pR5fgeQUE23kCQ1m7fOLbMmLFhgcCkvMgagmkDBounhqeR1nVYesgkPwsEJuVF1hAcP0aQBUCdzRwyKcACgUl5b/UIjh8sBmzA2KQECwQm5dW29ZKXmU5B1lvrKyvCPYJ6CwQmBVggMCnvWHsP04qyEZH+Y1medEpyMyw1ZFKCBQKT8ho6+qgoyD7huK0lMKnCAoFJea3dPkryMk44XlGYTV2H9QhM8rNAYFJeS7ePopwT9yeuKMiiwXoEJgW4GghEZKWI7BKRahG5dYhzLhKRTSKyTUSed7M9xgymqrT1eCnOPbFHUJyTQWuPLSgzyc+1MtQikg7cDlwK1ADrRORhVd0+4Jxi4OfASlU9JCIVbrXHmGi6vQF8AaU4J0ogyM2g2xvA6w+S6bHOs0lebv52LweqVXWfqnqB+4BrBp3zEeBBVT0EoKr1LrbHmBNEvvFH6xEU5YbSRW3WKzBJzs1AMBM4POBxTfjYQAuBEhF5TkQ2iMjHo72RiNwoIutFZH1DQ4NLzTWpqLXbCxB1jKAo3Eto6/GOaZuMGWtuBgKJcmxwKUcPcDZwJXAZ8P9EZOEJL1K9S1WrVLWqvLzc+ZaalNUWLipXMsQYAWCF50zSc3Oryhpg9oDHs4CjUc5pVNUuoEtEXgCWAbtdbJcx/d5KDZ3YI4ikiyw1ZJKdmz2CdcACEZkrIpnAKuDhQef8BXiHiHhEJBd4G7DDxTYZc5yWcGoo6hiB9QhMinCtR6CqfhG5BXgSSAfuUdVtIrI6/PydqrpDRJ4ANgNB4G5V3epWm4wZLHKTL4o2ayg8bmBTSE2yczM1hKquAdYMOnbnoMffB77vZjuMGUpbj4/sjDSyM9JPeK4g24OIpYZM8rPJ0SaltXZ7+7/5D5aWJhRmZ9DWbbOGTHKzQGBSWmu3L+r4QERxrq0uNsnPAoFJaa09wweCopwMSw2ZpGeBwKS04VJDEAoENmvIJDsLBCaljZwayrQegUl6FghMylJVWnt8FA2bGvJYIDBJzwKBSVm9viBef3DY1FBxTiat3V6CwcHVUYxJHhYITMpq7Rl6VXFEcW4GQYVOr3+smmXMmLNAYFJW6zAF5yIKIxVIbcDYJDELBCZlRQJBYZTyEhHFOVZ4ziQ/CwQmZbX3hgNB9vDrCMAKz5nkZoHApKyO3lDef7hAUGy7lJkUYIHApKzOcI8gP3vo2ouRgeRW26XMJDELBCZlRXoE+VlDB4KCcJCInGtMMrJAYFJWZ5+fLE8amZ6h/wxyMtJJTxM6ei01ZJKXBQKTstp7/RQMMz4AICLkZ3msR2CSmgUCk7I6+/z9qZ/hFGRbIDDJzQKBSVkdvb4YA0GGBQKT1CwQmJTV2esfdqA4ItQjsDECk7wsEJiU1dEbW2qo0FJDJslZIDApq7PPT37W8IPFEE4N9VmPwCQvVwOBiKwUkV0iUi0it0Z5/iIRaRORTeF/vulme4wZqD3mMQLrEZjkNvJfQYJEJB24HbgUqAHWicjDqrp90KkvqupVbrXDmGhUNeZZQ5Hpo6qKiIxB64wZW272CJYD1aq6T1W9wH3ANS5ez5iYdXkDqBLzrKFAUOnxBcagZcaMPTcDwUzg8IDHNeFjg50nIm+KyOMiclq0NxKRG0VkvYisb2hocKOtJsV09peXiGWMwHPca5z20Bs1/PsjgzvKxowdNwNBtD704P3+NgInqeoy4L+BP0d7I1W9S1WrVLWqvLzc2VaalBSZDhrrGAGEViI7LRhUfvjUbu55eT9vHGpx/P2NiYWbgaAGmD3g8Szg6MATVLVdVTvDP68BMkRkiottMgaAjr5wjyCm6aOhXoMbawnWHWimpqUHgDuf3+v4+xsTCzcDwTpggYjMFZFMYBXw8MATRGSahEffRGR5uD1NLrbJGGDgXgSx9wjcmDn04MYj5GWm88m3z+Wp7XXsbeh0/BrGjMS1QKCqfuAW4ElgB/BHVd0mIqtFZHX4tA8AW0XkTeCnwCpVHZw+MsZx8Y0RRHoEzgaCXl+ANVtquXzJdG66YB6q8Mz2OkevYUwsXJs+Cv3pnjWDjt054OefAT9zsw3GRBPPGEF+f4/A2dTQzmMddPT5uXTxVCoKsynLy2R/Y5ej1zAmFray2KSkzjjGCNxKDR1sCt3055fnATCvPI99FgjMOLBAYFJSZAZQfmYMPYJMDyJvDTA75UBjNyIwqyQXgLlT8qxHYMaFBQKTkiKVR9PSRl4pnJYm5Gc6X4H0YFMX0wuzyc5IB2DulHwaOvqs0qkZcxYITEqKdS+CCDfqDR1s7uaksrz+x3OnhH62XoEZaxYITEoKVR6NJxBkuNIjOKkst//xvHILBGZ8WCAwKSnWvQgi8h3uEXT0+mjs9B7XI5hTmosI7GuwQGDGlgUCk5I6+vzkj7Bx/UBOp4YONnUDUDmgR5Cdkc6skhzrEZgxZ4HApKT4xwgy+qecOiESCOYMCAQQGjC2QGDGmgUCk5I6e/0UxDVG4OysoYPNoZv9wNQQwNyyXA40WSAwY8sCgUlJ8Y4RFGR7HK0+erCxmyn5WScMWE8tyqaj10+P1/Y+MGPHAoFJOf5AkB5fIKY6QxGF2Rl4/UH6/M7coI+29TCzJOeE4+X5WQDUd/Q6ch1jYmGBwKScSK4/3h4BOFdmoqGjj4qCrBOOVxRmA1Df0efIdYyJhQUCk3IiN/NY6gxFRFI4TgaC8miBIHyswQKBGUMWCEzKiWcvgogCBzen8QWCNHV5+9NAA0UCQX27pYbM2LFAYFLOW6mh+NYRgDP7Fjd1egGi9ghKcjPxpImlhsyYskBgUk7kW318JSac27c4kvaJNkaQliZMyc+yQGDGlAUCk3ISGSx2ct/ihs5Q2idajwCgojDLxgjMmLJAYFJOewKDxU7OGorc5IcMBAXWIzBjywKBSTmd/YPFsY8RODlrqL49dJOfEmWwGKC8IJsGW0dgxpAFApNyOnp9eNKELE/sv/6e9DRyMtIdSg31UZjt6d+QZrCKgiyaurz4A8FRX8uYWFggMCmnsy9UXkJk5N3JBnKqAmlDR1//wrFoyguyUIXG8OwiY9xmgcCknI5ef1zjAxEF2R5HKpA2dPRFXUMQYYvKzFiLKRCIyAMicqWIxBU4RGSliOwSkWoRuXWY884RkYCIfCCe9zcmER29fgriqDMUUZCdQbtDqaGhBophYJkJGycwYyPWG/sdwEeAPSLyHRFZNNILRCQduB24HFgMXCcii4c477vAkzG32phR6Oj1JdwjGG1qSFWpbx8hEERWF1uPwIyRmAKBqj6jqn8HnAUcAJ4WkbUi8gkRGeqr1XKgWlX3qaoXuA+4Jsp5nwEeAOrjbr0xCejs88dVXiKi0IF9i7u8AXp8gWEDQWQ2kaWGzFiJOdUjImXADcA/AG8AtxEKDE8P8ZKZwOEBj2vCxwa+50zgfcCdI1z7RhFZLyLrGxoaYm2yMVF19Ma3cX2EEz2C4VYVR2R60ijI9tDcZYPFZmzEOkbwIPAikAtcrarvUdU/qOpngPyhXhblmA56/BPgq6o6bJF3Vb1LVatUtaq8vDyWJhszpNCsofjHCPKzRh8IGjtDgaBsmMFigNK8TAsEZszE+rXoblVdM/CAiGSpap+qVg3xmhpg9oDHs4Cjg86pAu4LT+ObAlwhIn5V/XOM7TImLqo6ijGCDHp8AXyBIBnpiU24i9zcy/Iyhz2vNC+Tlm4LBGZsxPrb/B9Rjr0ywmvWAQtEZK6IZAKrgIcHnqCqc1W1UlUrgfuBmy0IGDf1+YP4AhpXnaGIyGu6RjGFtCUcCEpGCARleZn9VUqNcduwfw0iMo1QXj9HRM7krXRPIaE00ZBU1S8itxCaDZQO3KOq20Rkdfj5YccFjHFDJLUTz8b1EQPrDRXnDn8jH0pz+Ft+6QivL8nNZOuR9oSuYUy8RvpruIzQAPEs4EcDjncAXx/pzcPppDWDjkUNAKp6w0jvZ8xoRWb9JDJGEHnNaNYSNHd6yclIJyczenmJiNL8TJq7vahq3CugjYnXsIFAVX8N/FpErlXVB8aoTca4JrIyOJFZQ4UOVCBt7vZSOkJaCEKpIa8/SJc3kFBbjYnHSKmhj6rq74BKEfmnwc+r6o+ivMyYCas/NZTAGEG+A4GgpctLSd7IvZGScOqoudNrgcC4bqTfsLzwv4eaImrMpJLIxvURTuxb3Nzt67/JD6csPxwIur3MKRt2OM6YURspNfSL8L//bWyaY4y7IjfxePYiiOjft3iUs4YqY7ix9/cIumx1sXFfrAvKvicihSKSISLPikijiHzU7cYZ47TRjBE4sUtZS1esYwShBWfNXaMvcmfMSGJdR/BuVW0HriK0UGwh8GXXWmWMS0aTGsrypJPpSUt41pDXH6Sjzz/i1FEIzRoC6xGYsRFrIIj0o68A7lXVZpfaY4yrOvv8ZGekJbwyuHAU9YYiK4VHWkwGkJeZTmZ6Gk1WZsKMgVi/Fj0iIjuBHuBmESkHrFi6mXQ6en0JrSGIKMjOSDgQRMpLxJIaEpFQmQkLBGYMxFqG+lbgPKBKVX1AF9FLShszoYU2pUl8Omao8FxiqaH+8hIxrkouscJzZozE8xdxKqH1BANf8xuH22OMqzp6/QmtIYgYTSnq/vISMfQIIFxvyAKBGQMx/UWIyG+B+cAmIFIyWrFAYCaZzr7E9iuOKMj20NTYndBrW+JIDUXOq2lJ7FrGxCPWv4gqYLGqDt5PwJhJpaPXR3l+4usjC0axS1lkKmhxbmxjFKXWIzBjJNapE1uBaW42xJix0Nk7+h5B4oPFfRRme2KesVSal0lHrx9fIJjQ9YyJVax/EVOA7SLyOtA/sVlV3+NKq4xxyejHCDLo9PoJBpW0tPiqgjZ3+2JOC8FbKaSWLi8VhdlxXcuYeMT6F/GvbjbCmLEQDCqd3tHNGirM9qAKnV5/3GUqQgXn4g8ETRYIjMti+otQ1edF5CRggao+IyK5hDabMWbS6PL6UU1sL4KISGmKjt74A0Fzl5fpRbHf0Af2CIxxU6y1hv6R0FaSvwgfmgn82aU2GeOK/jpDo0wNQWIVSFti3IsgYmCPwBg3xTpY/GngfKAdQFX3ABVuNcoYN4xmL4KI/gqkcQ4YqyrNMRaci+jvEdgm9sZlsQaCPlXt/20MLyqzqaRmUukvODeKMYJEK5B2ewP0+YNxjREU52Qggm1ib1wXayB4XkS+TmgT+0uBPwGPuNcsY5w3mv2KIxLdt7i/zlAcm9570tMoysmwMhPGdbEGgluBBmALcBOhDen/2a1GGeOGyBjBaFJDie5bHE/l0YFK8zL7S1MY45ZYi84FCQ0O36yqH1DVX8ayylhEVorILhGpFpFbozx/jYhsFpFNIrJeRN4e9ycwJkZOjBEkum/xW5VH4+uNlOVl0mypIeOyYQOBhPyriDQCO4FdItIgIt8c6Y1FJB24HbgcWAxcJyKLB532LLBMVc8A/h64O4HPYExMOh0YI8jJSCc9TeKeNdTfI4gjNRQ531JDxm0j9Qg+T2i20DmqWqaqpcDbgPNF5AsjvHY5UK2q+8IDzfcxqHS1qnYO6FnkYQPQxkUdvT5EIC8z8UAgIhRke+LetzhSZyiyBWWsyvItNWTcN1Ig+DhwnarujxxQ1X3AR8PPDWcmcHjA45rwseOIyPvCm948RqhXcAIRuTGcOlrf0NAwwmWNia6jz09+pifu0hCDJVJvqKXLS3qaxJ2WKskNbU5j9R6Nm0YKBBmq2jj4oKo28Nb2lUOJ9td2wm+zqj6kqouA9wLfivZGqnqXqlapalV5efkIlzUmutHWGYooyIq/AmlTl5eS3Iy4g1BpXib+oNKeYKE7Y2IxUiAYrk86Un+1Bpg94PEs4OhQJ6vqC8B8EZkywvsak5DRVh6NKMj2xH1jbunyxj0+AKHUEGDjBMZVIwWCZSLSHuWfDmDJCK9dBywQkbkikgmsAh4eeIKInCwiEv75LCATaErsoxgzvI6+0e1XHJHIvsXN3fEVnIuIBI/mrr4RzjQmccN+PVLVhAvLqapfRG4BniRUoO4eVd0mIqvDz98JXAt8XER8QA/wYdv8xrils9dPcQLfygcLjRHEOWuoy8v88vg3xIkMLkcGm41xw+j7ycNQ1TWEFp8NPHbngJ+/C3zXzTYYE9HR62d2ae6o3yehweJuL6X5CfQIwusOrEdg3BTrymJjJr32Xh+FOU6khkLTR2PtvAaDSku3L67yEhGRHoFVIDVuskBgUoKq0tbji3sPgWiKczIJBDXmtQTtvT4CQU1ojCAnM52cjHTbk8C4ygKBSQm9viC+gFLkQI+gKLz5fGt3bHn7RMtLRNgm9sZtFghMSmjrCd20nQgEkZk8sQaCRMtLRJTmZVqPwLjKAoFJCZGy0YU5o58fURzpEfTEdnOO7CcQz6Y0A5XmWb0h4y4LBCYlONkjKA6/R0uMPYJIWmdKfnx1hiIsNWTcZoHApIT2cCBwZLA4nOJpi7EY3FtjBJYaMhOTBQKTEpzsERTF2SNo7OwjP8tDdkZi6zNL8zLp8gbo9QUSer0xI7FAYFJCJBA4sY4g05NGfpYn5sHipk5vf82gRER6EjZOYNxigcCkhPae0Jz/QgeKzkGoVxDzYHFXH2UJpoXAAoFxnwUCkxLaenzkZabjSXfmV744NyPOHkFiA8VggcC4zwKBSQntvT5HxgciSnIzaY1xsLipy8sUB1JDLbZTmXGJBQKTEtp6nKkzFFEUY48gGFSau7xxb1E5UCSt1GSb2BuXWCAwKaHd4UBQkptBa8/IgaCtJ1RnKNGpoxCa8pqeJpYaMq6xQGBSQluPs6mh4pxQaigYHL4CaVO4fPRoZg2lpQkluRm2ib1xjQUCkxLaHao8GlGcm0FQoWOECqSRdE6iq4ojSnIzabbUkHGJBQKTEtp7/c72CPpXFw+fHoqUhhhNjwCs3pBxlwUCk/T8gSCdfX5HCs5FvFVvaPibc1NnODU0isFiCAUSSw0Zt1ggMEkvsq2ko9NH8yIVSIfvETR2RkpQj+7aJbnWIzDusUBgkl6bgwXnIopyInsSjNAj6OqjJDdj1AvZyvJCg9OBEQanjUmEBQKT9CJ7ETg7RhDbLmXNXaNbVRxRmpdJUN8KasY4yQKBSXr9lUdHmZ4ZKNYxgsZO76jqDEWU9JeZ6Bv1exkzmKuBQERWisguEakWkVujPP93IrI5/M9aEVnmZntManIjNeRJT6MoJ2PE1b4NHX1MKRh9jyAy2NzcZT0C4zzXAoGIpAO3A5cDi4HrRGTxoNP2Axeq6lLgW8BdbrXHpC4n9yIYqKIgi/qO3iGfV1WOtfUyvTB71NeKDE5bj8C4wc0ewXKgWlX3qaoXuA+4ZuAJqrpWVVvCD18FZrnYHpOiIrt7RW6mTqkozKKufegbc3uPnx5fgGlFow8EkR6BbVlp3OBmIJgJHB7wuCZ8bCifBB6P9oSI3Cgi60VkfUNDg4NNNKmguctHfpaHLE9iO4QNZWpBNg0dQweCY+2h3sJUB3sEtmWlcYObgUCiHIs6901ELiYUCL4a7XlVvUtVq1S1qry83MEmmlTQ3NXneG8AoLwwi4aOPlSjT+mMBAInegRZnnTyszzWIzCucDMQ1ACzBzyeBRwdfJKILAXuBq5R1SYX22NSVHO3j9Lc0c/cGayiIBtvIDjkFNK6tnAgcKBHALaJvXGPm4FgHbBAROaKSCawCnh44AkiMgd4EPiYqu52sS0mhbV0eUdVBnooFeHZQPVDpIciPYKKwtHPGoLQFFLrERg3uBYIVNUP3AI8CewA/qiq20RktYisDp/2TaAM+LmIbBKR9W61x6Su5i5v/zx8J70VCKLPHKpt66UsL9OxsYkyKzxnXOJcFa4oVHUNsGbQsTsH/PwPwD+42QZjmru87qSGwimf+iFmDtW19zoyUBxRmpfJztp2x97PmAhbWWySWo83QI8v4HKPYIjUUFuvIwPFEaXh1NBQg9PGJMoCgUlqkRIQTpR5GCwvy0NeZvqQqSE3egR9/iDd3oBj72kMWCAwSa65fzGZ84EAQumhaD2CPn+Api4v0x3uEQA2TmAcZ4HAJLXITdONWUMQSg81RBkjiIwbODV1FOgf57BAYJxmgcAktUhqqMSFwWII9QjqoqSG+lcVO9kjCG93aTuVGadZIDBJLVId1I0xAggXnms/cXXx0dYeAGdTQ5EegW1ibxxmgcAktZZuL2kChQ5XHo2YWphFjy9Ae4//uOPV9Z2kCZxUluvYtfp7BJYaMg6zQGCSWnOXl+LcTNLTopW+Gr15U/IB2NvYedzxPXWdVJblOVroriDLQ0a6WGrIOM4CgUlqLd3eUW8cP5wFU0OBYE9dx3HHqxs6mV+R7+i1RISyvKxhK54akwgLBCapNXV6+2v5u2FWSS5ZnjT21L3VI/AFghxo7GKBw4EAYHpxNrVtPY6/r0ltFghMUmvp9rpSgjoiPU2YX57Pnvq3AsHBpi78QeVkFwLBjKIcaluH3hXNmERYIDBJrdmlyqMDLZyaT/WAQBD5eUFFgePXmlaUzdG2HiszYRxlgcAkLV8gSFOXl/IC56ZwRrNgagFHWnvo7AvNHIoEgvkVeY5fa3pRNr2+ofdAMCYRFghM0grtHhaa4ummSAooEgD21HcysziH3Ezni/vOKM4BQiWujXGKBQKTtOrand0hbCiRQeHIzKFdxzpcGR+Atxao2YCxcZIFApO06hzcPH44J5XlkelJY3NNGztq29l5rIO3nzzFlWtFegRHrUdgHOTqxjTGjKe6cOE3twNBeppwxenT+OP6wxxt7SHLk8YHq2a5cq0p+Vl40oTaVusRGOdYj8AkrWPtvXjSxLU6QwN9eeUiAJ7dWc97z5hJsUtF7tLThKmF2TZGYBxlgSBBqsq+hk6OtfXaVL4Jqq69l4qCLNJcKi8x0MziHG68YB4AH19xkqvXmlGc3V/UzhgnWGooAYGg8tn73uCxzbUAXL1sBrd9+IwxueGMl7ZuH3lZ6XjSJ893h/r2vv59hcfC59+1kKuWzuCUac6vHxhoWlEObx5udfUaJrVYIEjANx7awmOba7npwnn4A8qvXtrPrJIcvhpODySTjYda+NIf32RfYxfzyvP49SeWM7vUuYqabjrW3svJ5e7M3okmPU1cDwIAM4qyeXJrL8GgJvWXDzN2XP16JyIrRWSXiFSLyK1Rnl8kIq+ISJ+IfMnNtjhl/YFm7lt3mJsunMfXLj+Vf77yVK5bPoc7ntubdN/S6tt7Wf3bDXgDQT53yQIaO/p438/XTpq0RGjPYHfXEIyH6UXZeANBGrus+JxxhmuBQETSgduBy4HFwHUisnjQac3AZ4EfuNUOp/30r9WU5WXyuUsWAKGKkN+48lQKsz38/LnqcW6ds77wx0109Pq5+/oqvnDpQv60egXtvT5++NTu8W7aiLq9fjp6/Y7uEDZRnDQltGL5YFN3Qq/v9QX49podfOjOV/j5c9W099oq5VTnZo9gOVCtqvtU1QvcB1wz8ARVrVfVdcCk+E1841ALL+xu4B8vmHfcqtH8LA/Xr6jkyW11VNd3DPMOk8e6A828XN3Ely87hUXTCgE4ZVoBN6yo5ME3ath5rH2cWzi8/qmjLpeXGA+RdNfe+s4RzjxRW4+P9/zsJX7xwj5aur1874ldfOp3GwgEbcJDKnMzEMwEDg94XBM+FjcRuVFE1ovI+oaGBkcal4hfrz1AYbaHj5574qyQG1ZUkp2Rxt0v7h+Hljnvzuf2UpqXyXXL5xx3/OaL5lOQ5eFHE7xXcKxtbBaTjYcZxTlkedLY2xB/IPjeEzupru/knhuqePqfLuR71y7l5eomfvz0xP7/adzlZiCINoqV0NcOVb1LVatUtaq8vHyUzUpMZ5+fJ7Yd4+plM8jPOnGMvSw/i6uWzuCxzbX0+gLj0ELn7K7r4Nmd9Vx/XiU5mcfvsFWcm8nHzjuJZ3bU9d9sJ6L68Iby04qSb4wgPU2YOyWPvQ1dcb1u/YFm/u+1Q3zi/Lm8c9FUAD50zmw+cPYsbn+umn0JBBaTHNwMBDXA7AGPZwFHXbyeqx7fUkuvL8j7zxp6xej7zpxJR5+fp7fXjWHLnHfv64fITE/j4+dFnw//oarZBBXu33A46vMTwdHW5O0RAMwvz4/7xv29J3cxvSibf7p04XHHv7pyEZnpadz5/F4nm2gmETcDwTpggYjMFZFMYBXwsIvXc9WDG49QWZbLWXOKhzzn3HllTCvM5qE3joxdwxwWCCqPbq7lolPKKRliRe5JZXmsmF/GH9YfJjhBc8sHm7oozcukINu9TWnG0/zyPA41d9Pnj633ufNYO6/vb+aGFZXkDerRlhdkseqc2Ty48QhHJsmMMOMs1wKBqvqBW4AngR3AH1V1m4isFpHVACIyTURqgH8C/llEakSk0K02JepYWy+v7GvifWfOQmToedvpacI1Z87g+d0NNHVOzql9r+5roqGjj2vOGH4458PnzOZwcw+v7msao5bFZ39jF5Vlk2O9QyLmV+QT1NhnDv3mlYNkedL4UNXsqM/feOF8IDQOZlKPq+sIVHWNqi5U1fmq+p/hY3eq6p3hn4+p6ixVLVTV4vDPE246ypPbjgFw5dLpI557zbKZBILKk9smZ3ro4U1HyctM55JTK4Y9792Lp5GXmc4jmydmtu9AUxeVU5zfGGaimB/HzKG2Hh8PbTzCe5bNGLKXN7M4h4sXVfDgxiP4A0FH22omvslTL2AcPb61lgUV+THVmD91egFzp+SxZkvtGLTMWX3+AGu21nLZadPIzkgf9tyczHQuXTyVx7cew+ufWDeObq+fuvY+5pYlbyCYGw5y+xpHHjB+YEMNPb4AHz+vctjzrj1rFo2dfby4p9GJJppJxALBCJo6+3h9fzOXnz4tpvNFhMtPn8Yr+5po7vK63DpnPbergY5eP+85Y0ZM51+9bAat3T5erp5YN44DjaF0STL3CPKyPMwoymbXseHXrQSDyu9ePciZc4pZMqto2HPfuaiCktwM7t9Y42RTx1WPN8Are5vYeKhl0s/mc5MFghE8tb2OoMLK00dOC0VcsWQ6gaDyVDil5BRfIMir+5p4ers7UzcffvMopXmZnB/jpirvWFBOYbZnwqWHDjaFviXPTeJAAHDGnGI2HGwZ9pyXqhvZ19jF9SP0BgAyPWlcc8ZMnt5eR1sS7Im8/kAzl9/2Atf98lXe//O1XHHbi+yonXCZ5wnBAsEIHt96jJPKcjl1euzFxE6bUcic0lzWbHUuEOxr6OTaO9ay6q5X+cffrOfiHzzH3S/uc6wEdmefn2e213HlkulkxFhhNNOTxsrTp/HUtroJ9W1rfzgQJHOPAGB5ZSlHWnuoaRl6wPg3rxygLC+Ty5fE1qO99qxZeP1BHt0ysYJ7vDYdbuUjv3yNgCq3f+Qsblt1Bp19ft7385fZdrRtvJs34VggGEZbt4+11Y2sPH3asLOFBhMRrlgynbXVjbR2jz49dLi5m/ffsZZDzd388IPLuH/1eayYX8Z/PLaD2//mTH2jp7cfo88f5JoY00IRVy+bQWefn+d2jd+K78EONHYxJT8r6sK/ZPK2eWUAvL6/Oerzh5u7eXZnPdctn0OWZ/gxn4jTZxaycGo+D2yYvOmhps4+PvW7DVQUZvHwp9/OlUunc80ZM3n0s2+nKCeDz/z+DTr7/OPdzAnFAsEwnt5Rhz+oXBFHWijiyiXT8QeVp0a5uKzXF+Cm324gGFQeuvl8rj17FlWVpdx9fRXvO3MmP3hqtyMpqL9sOsrM4hzOmlMS1+vOm1dGWV7mhEoPHWjsTuqpoxGnTC2gMNszZCD43WsHEeAjb5sT9floRIRrz5rFxkOtk3al8b88vI2mLi93fvTs42ZJVRRk89NVZ3KgqYv/eHT7OLZw4rFAMIwnttYysziHpSMMskVz+sxCZpXkjHr20I+f3s322nZ+suqM43LeIsK337+EpbOKuPXBLaPqeTSFZ4pcvWxG3PXtPelpXLFkOs/uqKNrgnzL2p/kU0cj0tKE5XNLowaCXl+AP647zLsXT+vf8D5W7ztzJmkCD0zCQeNX9zXx6OZaPnXhfE6feeLf7dvmlfHJt8/lD+sPs/WIpYgiLBAMobPPzwt7GrnstPjSQhEiwpVLpvNydWPCA2/7G7u45+X9fKhqVn9tmIGyM9L57rVLaevx8d0ndiZ0DYA1W48RCGrcaaGIq5ZOp9cX5Nmd9Qm3wSmNnX00dPSxIIapvslg+dxS9jV2Udt2/Irg3792iJZuH9evqIz7PSsKs3nHgnIe2nhkwq4cjyYQVP7tke3MLM5hdXiBXDSfuWQBJbmZfOvR7bbNbJgFgiE8GZ4ff+XS2AbZorly6XR8AeWxBHsF//nYdrI86XzpslOGPOfU6YV88u1zuff1w2w8NPwMkqE8vOkICyryWZTg7lrnVJYytTCLR94c//RQZHOgM2YXj2s7xsq7F08jTeB/B6wI7uzz87O/VXP+yWWcN78sofe99uxZHA2vqJ8sHnnzKDtq2/nq5YtOKJY4UGF2Bl+4dCGv7W/mrxPgy8tEYIFgCH958yizSuLPmQ+0ZGYRCyryEyrO9uKeBp7ZUc+nLz6ZihFq6n/ukgWUF2TxHwl8wznQ2MW6Ay2898yZCfV8IJSiuGrpDJ7f1UBbz/hOO9x0uJU0YcQ588mickoeVy2dwW9fOUhLeN3KHc9V09zl5SuXJb516rsXT6Ug2zNpBo19gSA/fmY3p04v5KolI4/prTpnNieV5fLDp3ZPql6PWywQRNHQ0cfL1Y1cc8aMhG+OEEoPfbAqNPAWT+14fyDIvz+ynTmlufz92ytHPD8vy8MXL13IxkOtrNkS38DxfesOk54mfODsoauqxuLqZTPwBoKOr52I16bDrSycWnDcxkHJ7pZ3nky3N8A3/ryFHz+9m9v/tpf3nzmTZaPoFWVnpHPV0hk8vvUYHZNgB7P7N9RwsKmbL166MKZxroz0ND7/rgVsr23niXH+nZ0ILBBE8djmowSCyntHKLwWi/eeMZP0NOH+OL5Z/d9rh9hT38k3rjw15ml/H6yazaJpBXzniR0xV6T0+oPcv+Ew71xUMepyzctmFTG7NIeHxzE9FAwqbx5u5cxhKsQmo4VTC7jpgnk8ua2O257dwxVLpvGda5eO+n1XnTObHl+ABzdO7Gq6vb4AP312D2fMLh6xRtZA71k2k5Mr8vnR07tTfoc2CwSDqCr3rTvMaTMKWTA1sZz5QBWF2Vx8SgV/WHeYHu/IN+jWbi8/fmY3K+aX8e7FJw4QDyU9Tfj6FadyuLmH36w9GNNrnt5eR2Onl48sj3164VBEhPeeMZOXqhs53JzYXrqjdaCpi/Zef8qMDwz0tStO5bWvX8Kvrq/ip6vOJNMz+j/tZbOLWTa7mF+/cmBCp0/uff0QtW29fOndp8TVg09PE/7p0oVU13dOiPGt8WSBYJANB1vYeawj6naUibrpwnk0d3n54/qRxwp+8swe2nt8fPPqxXGnpS5YWM6FC8v56V/3jFjnSFX5+XPVzCnN5YKFzuz6dt3yOQihHs142BQeKB5NSmQym5KfxSWnTsUT48rwWNyw4iT2NXTx0gSrJxXR1efn9r/t5dx5pZx/cvwD4ytPm8bi6YX85Jnd+FK46qoFgkF+++pBCrI8CU+ljOacylKqTirhrhf2DfvL9ubhVn776kGuWz6nf8P4eH3jylPp8Qb41ggLZp7aXse2o+189pIFpMe5dmAoM4pzuHTxVP6w7tC4lJx4qbqRwmwPCypG35MzIVcsmc6U/Ex+8ULiu5f1+gLc+/ohVv92Ayt/8gIfvHMt33l8J9X1wxfMi8Udz+2lsbOPr6xclNB4Xlqa8MV3L+RAUzf3vj4+X2AmAgsEA9S397JmSy3Xnj3L8cHGmy+ez5HWniE3/uj1Bfjin96kPD+Lr6xMfLbHwqkF3HzxyTz0xhH+ujP6qmZ/IMiPn97N3Cl5vNfBgAfw8fMqaen28ZdNY5tX7vMHeHpbHZedNs2xwGYgy5PO6gvn83J1U0JVZp/YeoyLf/AcX3twC1uOtDGrJLS47Zcv7uPSH7/ANx7akvBMs8PN3dz14j7ee8aMUc3ue+eiClbML+NHT+92pCTMZGSBYICf/a0aVfjE+ZWOv/fFp1TwrlMr+P6Tu6getJmIqvLNv2ylur6T731gKUU5o9te8ZaLT+aUqQV8+U+bORClXv1PntnDzmMdfPmyUxxNIwCsmF/G0llF/PTZ6pgHrZ3w4u5GOvr8MW0eZOLz0XNPYkZRNt97clfM05N9gSDfenQ7q3+3gdK8TP7vH97Gy7e+k7uvP4c/rV7Ba1+/hBtWVHLv64e4+r9fYvvR+KqCqir/+vA20gS+enniX5wgNL71zasX097j48dP7x7Ve01WFgjCDjeHuoYfOmc2J7mwoYmI8F/vX0JOZjqrf7ehv1SyLxDk3x7Zzh/X13DLxSc7kq/P9KRxx0fPIqjK9f/zOocGbGf4yJtHuf25aladM5srYphvHS8R4cuXncKR1h7uHcOxgjVbainKyYi5hLaJXXZGOp+/dCFvHm7lN6+MPBHhWFsvq+56lV+9tJ/rzzuJB29eccL/lyn5WfzL1afxp9Ur6PMHeP8dL/PnOPb6vvf1wzy7s56vXLaI6UXxldCIZtG0Qj5+XiW/efUga/dOzPEQN8lkW2JdVVWl69evd/x9P/37jTyzvY7nv3wx04pGN5VyOGv3NvKp320kqMq588rYUdtOTUsPN6yo5F8SGCAezsZDLXzs7tfwB5Wrls6gpdvLX3fWs2x2Mff947nDrr4cDVXlI798jV11HTz1hQuYkp/lynUiOnp9rPj2X7l8yTS+94Flrl4rVQWDyid/vY6Xq5t48OYVUev4ALxc3chn732DHl+A71y7lPcsGzn12NDRx6d/v5HX9zfzifMr+foVpw5bCn1zTSsf/sWrnH1SCb/5++Vx18caSrfXz1U/fYleX4DHP3cBRbmj65lPNCKyQVWroj5ngQD+sukIn7tvE1+8dCGfuWSBo+8dzaGmbr7zxA721ndRmOPh5otO5qJTyh0NAhG1bT3815qdvLK3CU+asGr5bG6+6GRHphcOZ9exDq7+2UtcsGAKv/x4lSufLeK7T+zkjuf28sgtb0+ZFcXjobnLy+W3vYA/oNz5sbM5p7K0/7mOXh8/fGo3v37lACeX53PHR8/i5DgG7X2BIN9es5N7Xt7P8spSfvDBZcyJUkF2R2071/3yVfKzPDzwqRWjXv8y2KbDrXzwzrUsm1XMbz/5tri+LKkq2462s/1oO0fbeggGlaLcTOZNyeOsk0pGnfIdLQsEw9hd18G1P1/LKdMK+MNN59lAo4N+9dJ+vvXodr66chGfumjoImCjUdPSzTt/+DxXLZnOjz58hivXMG/Z29DJP/x6PYebu3nnogoWzyikpqWHJ7Yeo8vr5/rzKvnyZaeQl+BeEH9+4wjfeGgL/qByw/mVfPDs2cydkkdDRx8PbKzhtmf3UJKbwZ9uWhE1UDjhsc213HLvRlbML+Onq86kbJgebSCobDzUwuNbjvHktmMcae2Jel6ahGYPfuDsWVyxZHrC/31GwwLBELYfbeejv3oNT5rwwKdWMLs0+WvYj6VgUPnMfW/w2OZabr18ETddMM/RnkFnn5/r73mdrUfa+NuXLoq73LJJTGu3l5/9tZqH3zxKfUcfBdke3r14GjesqHSkR1bb1sN/PraDNVtqCWroJhpZz7bytGn8+zWnUeFwT2Cw+zfU8PWHtlCck8FnLlnANWfMoDA79I2+vdfHxoMtPL29jie31dHY2UemJ40LFkzhstOmcU5lKbNKckhPE5q7vOyu62Tt3kYeefMoB5q6yc1M54ol0/lQ1WzOqSxxtbc80LgFAhFZCdwGpAN3q+p3Bj0v4eevALqBG1R143Dv6UQg6PEG+J+1+/nJM3sozc3k3hvPTfr9bceLPxDk83/YxKOba7lwYTn/fOWpjqzY3t/YxZf+9CabDrfys+vO5HIXBr7N8FSVQFAdn3kWUdfeyzM76qhr66UwJ4MV86eweEZi62sSsf1oO199YDNbwvsWlBdkEQgqLd1eVCE3M52LT6lg5enTuHhRxYg74qkqGw62cP+GGh7dXEtnn5/Kslzee+ZMLlxYztJZxa5mJMYlEIhIOrAbuBSoAdYB16nq9gHnXAF8hlAgeBtwm6q+bbj3TTQQ1LX38vyuBl7b38xT247R0efn8tOn8W/vcf/bRaoLBpXfvnqQ7z6xk25vgLPmFLNi/hROrsinvCCL8oIscjPTSU8T0iTyD3gDQbz+IH3+IH2+IC3dXvY3dvFydSPP7qwny5PG9z+wzKaMGteoKm8cbuXlPY0cau4m05PGtMJszpxTQlVlCdkZiU246Pb6eXzLMf6w/jDrDjSjCsW5GZxTWcrJFfnML89ndkkOxbmZFOdmUJDtIcuTPqpAMV6B4DzgX1X1svDjrwGo6rcHnPML4DlVvTf8eBdwkaoOWcA/0UDw2OZaPv37jZTkZnDRKRV85G1zjhvsMu5r7Ozj/g01PLa5lu217QkX+ppZnMNlp03jUxfNp7zA3RlJxritucvLi3saeGF3I5sOt3CwqRv/EH8bN104j69dfmpC1xkuELg5YjETGFhcp4bQt/6RzpkJHBcIRORG4Mbww85wwIhmCjDsJOCDwCbgJ8OdNPGN+DmTyAmf9SCwFviXcWmOa1L6/2mScvxzfv278PXEXz5kATU3A0G0PszgMBfLOajqXcBdI15QZP1QES+ZpMrnhNT5rKnyOSF1Putk+pxuTiavAWYPeDwLGFzrNZZzjDHGuMjNQLAOWCAic0UkE1gFPDzonIeBj0vIuUDbcOMDxhhjnOdaakhV/SJyC/Akoemj96jqNhFZHX7+TmANoRlD1YSmj35ilJcdMX2UJFLlc0LqfNZU+ZyQOp910nzOSbegzBhjjLOs+qgxxqQ4CwTGGJPikiIQiMhKEdklItUicut4t8ctIjJbRP4mIjtEZJuIfG682+QmEUkXkTdE5NHxboubRKRYRO4XkZ3h/7fnjXeb3CAiXwj/3m4VkXtFJGmW9IvIPSJSLyJbBxwrFZGnRWRP+N+Jb6PmskkfCMKlLG4HLgcWA9eJyOLxbZVr/MAXVfVU4Fzg00n8WQE+B+wY70aMgduAJ1R1EbCMJPzMIjIT+CxQpaqnE5pAsmp8W+Wo/wVWDjp2K/Csqi4Ang0/npAmfSAAlgPVqrpPVb3AfcA149wmV6hqbaQon6p2ELphzBzfVrlDRGYBVwJ3j3db3CQihcAFwK8AVNWrqq3j2ij3eIAcEfEAuSTRmiFVfQFoHnT4GuDX4Z9/Dbx3LNsUj2QIBEOVqUhqIlIJnAm8Ns5NcctPgK8AwXFuh9vmAQ3A/4TTYHeLSNKVwlXVI8APgEOESsi0qepT49sq102NrIsK/7tinNszpGQIBDGVqUgmIpIPPAB8XlXj2/V7EhCRq4B6Vd0w3m0ZAx7gLOAOVT0T6GICpxASFc6PXwPMBWYAeSLy0fFtlYlIhkCQUmUqRCSDUBD4P1V9cLzb45LzgfeIyAFCqb53isjvxrdJrqkBalQ10rO7n1BgSDbvAvaraoOq+oAHgRXj3Ca31YnIdIDwv+vHuT1DSoZAEEspi6QQ3sjnV8AOVf3ReLfHLar6NVWdpaqVhP5//lVVk/Lbo6oeAw6LyCnhQ5cA24d5yWR1CDhXRHLDv8eXkISD4oM8DFwf/vl64C/j2JZhjf3GmQ4bqpTFODfLLecDHwO2iMim8LGvq+qa8WuSccBngP8Lf5HZx+hLrUw4qvqaiNwPbCQ0++0NJlEJhpGIyL3ARcAUEakhVCX9O8AfReSThALhB8evhcOzEhPGGJPikiE1ZIwxZhQsEBhjTIqzQGCMMSnOAoExxqQ4CwTGGJPiLBAYY0yKs0BgzBgIF1ozZkKyQGDMEEQkT0QeE5E3wzX0Pywi54jI2vCx10WkQESyReR/RGRLuHDcxeHX3yAifxKRR4Cnwu93j4isC5+XlFVyzeRj31KMGdpK4KiqXgkgIkWEVsR+WFXXhUtI9xDaNwFVXSIiiwjd9BeG3+M8YKmqNovIfxEql/H3IlIMvC4iz6hq1xh/LmOOYz0CY4a2BXiXiHxXRN4BzAFqVXUdgKq2q6ofeDvw2/CxncBBIBIInlbVSJ36dwO3hsuDPAdkh9/TmHFlPQJjhqCqu0XkbOAK4NvAU0QvcR6tFHrEwG/7Alyrqruca6Uxo2c9AmOGICIzgG5V/R2hTVXOBWaIyDnh5wvCg8AvAH8XPraQ0Lf8aDf7J4HPhKtvIiJnuv8pjBmZ9QiMGdoS4PsiEgR8wKcIfav/bxHJITQ+8C7g58CdIrKFUGXNG1S1L3y/H+hbhHZe2xwOBgeAq8bgcxgzLKs+aowxKc5SQ8YYk+IsEBhjTIqzQGCMMSnOAoExxqQ4CwTGGJPiLBAYY0yKs0BgjDEp7v8DbZRvu+zcPKAAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Kernel density estimate\n",
    "ax = sns.kdeplot(ve_df.score)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "3a0b00b4",
   "metadata": {},
   "source": [
    "We see that score exhibits a multimodal distribution. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 163,
   "id": "64accd96",
   "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": [
    "# violin plot\n",
    "ax = sns.violinplot(x=ve_df.score)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "9f3e5d29",
   "metadata": {},
   "source": [
    "The violinplot shows that the majority of the scores are either 5 or 6. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 164,
   "id": "30833ee0",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<AxesSubplot:xlabel='score', ylabel='fuelType1'>"
      ]
     },
     "execution_count": 164,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "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.boxplot(x='score', y='fuelType1', data=ve_df)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "306ab1a6",
   "metadata": {},
   "source": [
    "From the above boxplot, we see that Diesel cars have lower scores without any outliers. For Gasoline cars, the median score is 5. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 165,
   "id": "4392ec19",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<AxesSubplot:xlabel='score', ylabel='year'>"
      ]
     },
     "execution_count": 165,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "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.boxplot(x='score', y='year', data=ve_df)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "427994d1",
   "metadata": {},
   "source": [
    "From this boxplot, we notice that as we move ahead with time, the emission scores increase. This is a realistic observation as newer car makes have better fuel efficiency and emissions standards. Therefore, we should treat year as a numeric variable and not categorical."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 166,
   "id": "8c3578e4",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<AxesSubplot:xlabel='VClass', ylabel='score'>"
      ]
     },
     "execution_count": 166,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x216 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.rcParams[\"figure.figsize\"] = (10,3)\n",
    "sns.boxplot(x='VClass', y='score', data=ve_df)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "56f13503",
   "metadata": {},
   "source": [
    "We see that there is very little score variation between the vehicle classes. Thus, this variable can also be dropped.  "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 167,
   "id": "dbef9aa1",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x216 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "ax = sns.barplot(x=\"score\", y=\"barrels08\", ci='sd', data=ve_df, capsize=.2)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "d32e469d",
   "metadata": {},
   "source": [
    "The score vs barrels08 barplot shows that the higher the score is, the annual fuel consumption (in barrels) is lower. Noticeably, green vehicles, have the least consumption. \n",
    "\n",
    "The correlation and distribution analysis suggests removing phevBlended and VClass.So, we remove them."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 168,
   "id": "31a6d1a5",
   "metadata": {},
   "outputs": [],
   "source": [
    "ve_df = ve_df.drop(columns=['phevBlended', 'VClass'])"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a7e27d71",
   "metadata": {},
   "source": [
    "### 3.3.4 Outlier detection using Boxplot rule"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 169,
   "id": "20160004",
   "metadata": {},
   "outputs": [],
   "source": [
    "q1 = ve_df.score.quantile(0.25)\n",
    "q3 = ve_df.score.quantile(0.75)\n",
    "\n",
    "# Calculate the interquantile range (IQR)\n",
    "IQR = q3 - q1\n",
    "ll = q1 - 1.5 * IQR\n",
    "ul = q3 + 1.5 * IQR"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 170,
   "id": "b3312835",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Lower limit outliers: 5015\n",
      "Upper limit outliers: 2031\n",
      "Total outliers: 7046\n"
     ]
    }
   ],
   "source": [
    "ll_outliers = ve_df.score < ll\n",
    "ul_outliers = ve_df.score > ul\n",
    "print('Lower limit outliers:', ll_outliers.sum())\n",
    "print('Upper limit outliers:', ul_outliers.sum())\n",
    "print('Total outliers:', ll_outliers.sum()  + ul_outliers.sum())"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f8bbf6c2",
   "metadata": {},
   "source": [
    "Although we detect 7046 outliers, we are not removing these at this moment. Since there are a sizeable number of outliers, removing them will reduce the data set size. We can use ensemble machine learning algorithms such as Random Forests which are somewhat robust to outliers. Later on, if we do no get satisfactory model performance, we can remove these outliers."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 171,
   "id": "0285c684",
   "metadata": {},
   "outputs": [],
   "source": [
    "ve_df.to_csv('data/VE_Cleaned.csv', index=False)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "e0d21ec2",
   "metadata": {},
   "source": [
    "### 3.3.5 Principal Component Analysis"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 172,
   "id": "90f8b704",
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "from sklearn.decomposition import PCA\n",
    "from sklearn.preprocessing import RobustScaler"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 173,
   "id": "68cc7f16",
   "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>barrels08</th>\n",
       "      <th>city08</th>\n",
       "      <th>comb08</th>\n",
       "      <th>fuelCost08</th>\n",
       "      <th>fuelType1</th>\n",
       "      <th>highway08</th>\n",
       "      <th>year</th>\n",
       "      <th>score</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>18.311667</td>\n",
       "      <td>16.0</td>\n",
       "      <td>18.0</td>\n",
       "      <td>3200.0</td>\n",
       "      <td>Premium Gasoline</td>\n",
       "      <td>22.0</td>\n",
       "      <td>2004</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>15.695714</td>\n",
       "      <td>18.0</td>\n",
       "      <td>21.0</td>\n",
       "      <td>2750.0</td>\n",
       "      <td>Premium Gasoline</td>\n",
       "      <td>26.0</td>\n",
       "      <td>2004</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>15.695714</td>\n",
       "      <td>18.0</td>\n",
       "      <td>21.0</td>\n",
       "      <td>2750.0</td>\n",
       "      <td>Premium Gasoline</td>\n",
       "      <td>26.0</td>\n",
       "      <td>2004</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>15.695714</td>\n",
       "      <td>19.0</td>\n",
       "      <td>21.0</td>\n",
       "      <td>2750.0</td>\n",
       "      <td>Premium Gasoline</td>\n",
       "      <td>25.0</td>\n",
       "      <td>2004</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>15.695714</td>\n",
       "      <td>18.0</td>\n",
       "      <td>21.0</td>\n",
       "      <td>2750.0</td>\n",
       "      <td>Premium Gasoline</td>\n",
       "      <td>26.0</td>\n",
       "      <td>2004</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   barrels08  city08  comb08  fuelCost08         fuelType1  highway08  year  \\\n",
       "0  18.311667    16.0    18.0      3200.0  Premium Gasoline       22.0  2004   \n",
       "1  15.695714    18.0    21.0      2750.0  Premium Gasoline       26.0  2004   \n",
       "2  15.695714    18.0    21.0      2750.0  Premium Gasoline       26.0  2004   \n",
       "3  15.695714    19.0    21.0      2750.0  Premium Gasoline       25.0  2004   \n",
       "4  15.695714    18.0    21.0      2750.0  Premium Gasoline       26.0  2004   \n",
       "\n",
       "   score  \n",
       "0      1  \n",
       "1      1  \n",
       "2      1  \n",
       "3      1  \n",
       "4      1  "
      ]
     },
     "execution_count": 173,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "ve_df = pd.read_csv('data/VE_Cleaned.csv')\n",
    "ve_df.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "56e92791",
   "metadata": {},
   "source": [
    "#### Data Normalization"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 174,
   "id": "5a1b6a34",
   "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>count</th>\n",
       "      <th>mean</th>\n",
       "      <th>std</th>\n",
       "      <th>min</th>\n",
       "      <th>25%</th>\n",
       "      <th>50%</th>\n",
       "      <th>75%</th>\n",
       "      <th>max</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>28556.0</td>\n",
       "      <td>-0.000854</td>\n",
       "      <td>0.816061</td>\n",
       "      <td>-3.049522</td>\n",
       "      <td>-0.489796</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.510204</td>\n",
       "      <td>4.081633</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>28556.0</td>\n",
       "      <td>0.213108</td>\n",
       "      <td>1.704423</td>\n",
       "      <td>-1.833333</td>\n",
       "      <td>-0.500000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.500000</td>\n",
       "      <td>21.833333</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>28556.0</td>\n",
       "      <td>0.248414</td>\n",
       "      <td>1.367579</td>\n",
       "      <td>-1.714286</td>\n",
       "      <td>-0.428571</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.571429</td>\n",
       "      <td>17.285714</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>28556.0</td>\n",
       "      <td>-0.001009</td>\n",
       "      <td>0.824696</td>\n",
       "      <td>-2.411765</td>\n",
       "      <td>-0.529412</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.470588</td>\n",
       "      <td>4.647059</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>28556.0</td>\n",
       "      <td>0.128537</td>\n",
       "      <td>1.123043</td>\n",
       "      <td>-1.875000</td>\n",
       "      <td>-0.500000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.500000</td>\n",
       "      <td>13.375000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>28556.0</td>\n",
       "      <td>-0.053592</td>\n",
       "      <td>0.606328</td>\n",
       "      <td>-1.125000</td>\n",
       "      <td>-0.625000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.375000</td>\n",
       "      <td>1.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>28556.0</td>\n",
       "      <td>0.160527</td>\n",
       "      <td>1.823389</td>\n",
       "      <td>-4.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>5.000000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "     count      mean       std       min       25%  50%       75%        max\n",
       "0  28556.0 -0.000854  0.816061 -3.049522 -0.489796  0.0  0.510204   4.081633\n",
       "1  28556.0  0.213108  1.704423 -1.833333 -0.500000  0.0  0.500000  21.833333\n",
       "2  28556.0  0.248414  1.367579 -1.714286 -0.428571  0.0  0.571429  17.285714\n",
       "3  28556.0 -0.001009  0.824696 -2.411765 -0.529412  0.0  0.470588   4.647059\n",
       "4  28556.0  0.128537  1.123043 -1.875000 -0.500000  0.0  0.500000  13.375000\n",
       "5  28556.0 -0.053592  0.606328 -1.125000 -0.625000  0.0  0.375000   1.000000\n",
       "6  28556.0  0.160527  1.823389 -4.000000  0.000000  0.0  1.000000   5.000000"
      ]
     },
     "execution_count": 174,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Data normalization\n",
    "robust_scaler = RobustScaler()\n",
    "ve_norm_arr = robust_scaler.fit_transform(ve_df.loc[:, ve_df.columns != 'fuelType1'])\n",
    "ve_norm_df = pd.DataFrame(ve_norm_arr)\n",
    "ve_norm_df.describe().T"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "217d018f",
   "metadata": {},
   "source": [
    "#### Full-component PCA solution"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 175,
   "id": "c2255d0b",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "PCA(n_components=7)"
      ]
     },
     "execution_count": 175,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Fit linear Kernel\n",
    "pca_linear = PCA(n_components=7)\n",
    "pca_linear.fit(ve_norm_arr)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 176,
   "id": "0fd570ed",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[ 69.09  91.27  95.99  98.99  99.59  99.98 100.  ]\n"
     ]
    }
   ],
   "source": [
    "#Cumulative Variance\n",
    "var_pca = np.cumsum(np.round(pca_linear.explained_variance_ratio_, decimals=4) * 100)\n",
    "print(var_pca)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 177,
   "id": "a1829912",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x216 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Scree plot\n",
    "var_pca = pd.DataFrame(var_pca, index=np.arange(1, 8))\n",
    "plt.plot(var_pca, color='blue')\n",
    "plt.title('Scree Plot')\n",
    "plt.xlabel('Number of Principal Components')\n",
    "plt.ylabel('Cumulative Variance Explained (%)')\n",
    "plt.savefig('scree_plot.png', dpi=600, bbox_inches='tight')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "aee91f19",
   "metadata": {},
   "source": [
    "From the cummulative variance data and the screen plot obtained using linear PCA, we find that 3 components are sufficient to explain 96% of the variance. Therefore, 3 orthogonal (uncorrelated) components can keep more than 95% variance of the 7 variables in the original dataset. This implies we would be obtaining a good solution."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 178,
   "id": "fe9a085c",
   "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>PC1</th>\n",
       "      <th>PC2</th>\n",
       "      <th>PC3</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>-3.078966</td>\n",
       "      <td>3.256219</td>\n",
       "      <td>0.326566</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>-2.244583</td>\n",
       "      <td>3.593229</td>\n",
       "      <td>-0.161929</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>-2.244583</td>\n",
       "      <td>3.593229</td>\n",
       "      <td>-0.161929</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>-2.195739</td>\n",
       "      <td>3.621737</td>\n",
       "      <td>-0.102962</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>-2.244583</td>\n",
       "      <td>3.593229</td>\n",
       "      <td>-0.161929</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>-2.196217</td>\n",
       "      <td>3.614991</td>\n",
       "      <td>-0.163369</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>-1.782508</td>\n",
       "      <td>2.710368</td>\n",
       "      <td>-0.070660</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>-2.534830</td>\n",
       "      <td>3.472266</td>\n",
       "      <td>-0.028479</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>-2.121121</td>\n",
       "      <td>2.567644</td>\n",
       "      <td>0.064230</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>-3.127811</td>\n",
       "      <td>3.227711</td>\n",
       "      <td>0.267599</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "        PC1       PC2       PC3\n",
       "0 -3.078966  3.256219  0.326566\n",
       "1 -2.244583  3.593229 -0.161929\n",
       "2 -2.244583  3.593229 -0.161929\n",
       "3 -2.195739  3.621737 -0.102962\n",
       "4 -2.244583  3.593229 -0.161929\n",
       "5 -2.196217  3.614991 -0.163369\n",
       "6 -1.782508  2.710368 -0.070660\n",
       "7 -2.534830  3.472266 -0.028479\n",
       "8 -2.121121  2.567644  0.064230\n",
       "9 -3.127811  3.227711  0.267599"
      ]
     },
     "execution_count": 178,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pca_linear = PCA(n_components=3)\n",
    "ve_pca_transform_arr = pca_linear.fit_transform(ve_norm_arr)\n",
    "ve_pca_transform_df = pd.DataFrame(ve_pca_transform_arr)\n",
    "ve_pca_transform_df.columns = ['PC1', 'PC2', 'PC3']\n",
    "ve_pca_transform_df.head(10)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 179,
   "id": "ddbf600c",
   "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>0</th>\n",
       "      <th>1</th>\n",
       "      <th>2</th>\n",
       "      <th>3</th>\n",
       "      <th>4</th>\n",
       "      <th>5</th>\n",
       "      <th>6</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1.000000</td>\n",
       "      <td>-0.736932</td>\n",
       "      <td>-0.780911</td>\n",
       "      <td>0.910707</td>\n",
       "      <td>-0.818702</td>\n",
       "      <td>-0.294785</td>\n",
       "      <td>-0.425487</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>-0.736932</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.993217</td>\n",
       "      <td>-0.648146</td>\n",
       "      <td>0.955490</td>\n",
       "      <td>0.225552</td>\n",
       "      <td>0.383834</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>-0.780911</td>\n",
       "      <td>0.993217</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>-0.690899</td>\n",
       "      <td>0.980921</td>\n",
       "      <td>0.240110</td>\n",
       "      <td>0.396941</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>0.910707</td>\n",
       "      <td>-0.648146</td>\n",
       "      <td>-0.690899</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>-0.726165</td>\n",
       "      <td>-0.216122</td>\n",
       "      <td>-0.385754</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>-0.818702</td>\n",
       "      <td>0.955490</td>\n",
       "      <td>0.980921</td>\n",
       "      <td>-0.726165</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.252753</td>\n",
       "      <td>0.408956</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>-0.294785</td>\n",
       "      <td>0.225552</td>\n",
       "      <td>0.240110</td>\n",
       "      <td>-0.216122</td>\n",
       "      <td>0.252753</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.231791</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>-0.425487</td>\n",
       "      <td>0.383834</td>\n",
       "      <td>0.396941</td>\n",
       "      <td>-0.385754</td>\n",
       "      <td>0.408956</td>\n",
       "      <td>0.231791</td>\n",
       "      <td>1.000000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "          0         1         2         3         4         5         6\n",
       "0  1.000000 -0.736932 -0.780911  0.910707 -0.818702 -0.294785 -0.425487\n",
       "1 -0.736932  1.000000  0.993217 -0.648146  0.955490  0.225552  0.383834\n",
       "2 -0.780911  0.993217  1.000000 -0.690899  0.980921  0.240110  0.396941\n",
       "3  0.910707 -0.648146 -0.690899  1.000000 -0.726165 -0.216122 -0.385754\n",
       "4 -0.818702  0.955490  0.980921 -0.726165  1.000000  0.252753  0.408956\n",
       "5 -0.294785  0.225552  0.240110 -0.216122  0.252753  1.000000  0.231791\n",
       "6 -0.425487  0.383834  0.396941 -0.385754  0.408956  0.231791  1.000000"
      ]
     },
     "execution_count": 179,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "ve_norm_df.corr()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 180,
   "id": "023b29f6",
   "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>PC1</th>\n",
       "      <th>PC2</th>\n",
       "      <th>PC3</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>PC1</th>\n",
       "      <td>1.000000e+00</td>\n",
       "      <td>1.245496e-15</td>\n",
       "      <td>-8.666262e-16</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>PC2</th>\n",
       "      <td>1.245496e-15</td>\n",
       "      <td>1.000000e+00</td>\n",
       "      <td>4.141852e-16</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>PC3</th>\n",
       "      <td>-8.666262e-16</td>\n",
       "      <td>4.141852e-16</td>\n",
       "      <td>1.000000e+00</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "              PC1           PC2           PC3\n",
       "PC1  1.000000e+00  1.245496e-15 -8.666262e-16\n",
       "PC2  1.245496e-15  1.000000e+00  4.141852e-16\n",
       "PC3 -8.666262e-16  4.141852e-16  1.000000e+00"
      ]
     },
     "execution_count": 180,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "ve_pca_transform_df.corr()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "12e55ed5",
   "metadata": {},
   "source": [
    "When we compare the normalized data correlation and the PCA transformed correlation, we see that the principal components are almost uncorrelated with each other. Earlier, e.g, city08 (city MPG for fuelType1) had a high correlation with comb08 (combined MPG for fuelType1)."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 181,
   "id": "cad7ccb8",
   "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>PC1</th>\n",
       "      <th>PC2</th>\n",
       "      <th>PC3</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>barrels08</th>\n",
       "      <td>-0.244742</td>\n",
       "      <td>-0.051528</td>\n",
       "      <td>0.571935</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>city08</th>\n",
       "      <td>0.583264</td>\n",
       "      <td>0.301618</td>\n",
       "      <td>0.345153</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>comb08</th>\n",
       "      <td>0.474396</td>\n",
       "      <td>0.233715</td>\n",
       "      <td>0.152760</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>fuelCost08</th>\n",
       "      <td>-0.223486</td>\n",
       "      <td>-0.043389</td>\n",
       "      <td>0.701621</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>highway08</th>\n",
       "      <td>0.386928</td>\n",
       "      <td>0.174093</td>\n",
       "      <td>-0.011527</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>year</th>\n",
       "      <td>0.063542</td>\n",
       "      <td>-0.034865</td>\n",
       "      <td>-0.171534</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>score</th>\n",
       "      <td>0.413709</td>\n",
       "      <td>-0.904623</td>\n",
       "      <td>0.092710</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                 PC1       PC2       PC3\n",
       "barrels08  -0.244742 -0.051528  0.571935\n",
       "city08      0.583264  0.301618  0.345153\n",
       "comb08      0.474396  0.233715  0.152760\n",
       "fuelCost08 -0.223486 -0.043389  0.701621\n",
       "highway08   0.386928  0.174093 -0.011527\n",
       "year        0.063542 -0.034865 -0.171534\n",
       "score       0.413709 -0.904623  0.092710"
      ]
     },
     "execution_count": 181,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# PCA loadings\n",
    "pd.DataFrame(pca_linear.components_.transpose(),\n",
    "             index=ve_df.loc[:, ve_df.columns != 'fuelType1'].columns,\n",
    "             columns=ve_pca_transform_df.columns)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "5c22d4c5",
   "metadata": {},
   "source": [
    "#### Visualizing PCA Components"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 189,
   "id": "224b1f2b",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 480x480 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "%matplotlib widget\n",
    "%matplotlib inline\n",
    "\n",
    "ve_df.score = pd.Categorical(ve_df.score)\n",
    "score_color = ve_df.score.cat.codes\n",
    " \n",
    "# Plot initialisation\n",
    "fig = plt.figure(figsize=(480/96, 480/96), dpi=96)\n",
    "ax = fig.add_subplot(111, projection='3d')\n",
    "ax.scatter(ve_pca_transform_df['PC1'], ve_pca_transform_df['PC2'], ve_pca_transform_df['PC3'], \n",
    "           c=score_color, cmap=\"tab20c\", s=60) \n",
    "# make simple, bare axis lines through space:\n",
    "xAxisLine = ((min(ve_pca_transform_df['PC1']), max(ve_pca_transform_df['PC1'])), (0, 0), (0,0))\n",
    "ax.plot(xAxisLine[0], xAxisLine[1], xAxisLine[2], 'r')\n",
    "yAxisLine = ((0, 0), (min(ve_pca_transform_df['PC2']), max(ve_pca_transform_df['PC2'])), (0,0))\n",
    "ax.plot(yAxisLine[0], yAxisLine[1], yAxisLine[2], 'r')\n",
    "zAxisLine = ((0, 0), (0,0), (min(ve_pca_transform_df['PC3']), max(ve_pca_transform_df['PC3'])))\n",
    "ax.plot(zAxisLine[0], zAxisLine[1], zAxisLine[2], 'r')\n",
    " \n",
    "# label the axes\n",
    "ax.set_xlabel(\"PC1\")\n",
    "ax.set_ylabel(\"PC2\")\n",
    "ax.set_zlabel(\"PC3\")\n",
    "plt.title(\"Total Explained Variance: {:.2f}%\".format(pca_linear.explained_variance_ratio_.sum() * 100))\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "54e88952",
   "metadata": {},
   "source": [
    "Note that for PCA analysis, we leave out fuelType1 even though it is an important variable."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "1de8ee02",
   "metadata": {},
   "source": [
    "# 4. Predictive Modeling"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "1901a66a",
   "metadata": {},
   "source": [
    "## 4.1 Loading cleaned data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "6a0c9076",
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "d9936d1c",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Note that we are not using dimension reduced data sets from PCA or correlation analysis.\n",
    "ve_df_all = pd.read_csv('data/Vehicle_Emissions.csv')\n",
    "ve_df = pd.read_csv('data/VE_Cleaned.csv')\n",
    "ve_df['VClass'] = ve_df_all['VClass'] # Getting original VClass"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "2f3386cf",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Creating dummies\n",
    "\n",
    "ve_df = pd.get_dummies(ve_df, columns=['fuelType1', 'VClass'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "b71ea112",
   "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>barrels08</th>\n",
       "      <th>city08</th>\n",
       "      <th>comb08</th>\n",
       "      <th>fuelCost08</th>\n",
       "      <th>highway08</th>\n",
       "      <th>year</th>\n",
       "      <th>score</th>\n",
       "      <th>fuelType1_Diesel</th>\n",
       "      <th>fuelType1_Green Fuel</th>\n",
       "      <th>fuelType1_Premium Gasoline</th>\n",
       "      <th>...</th>\n",
       "      <th>VClass_Sport Utility Vehicle - 2WD</th>\n",
       "      <th>VClass_Sport Utility Vehicle - 4WD</th>\n",
       "      <th>VClass_Standard Pickup Trucks 2WD</th>\n",
       "      <th>VClass_Standard Pickup Trucks 4WD</th>\n",
       "      <th>VClass_Standard Sport Utility Vehicle 2WD</th>\n",
       "      <th>VClass_Standard Sport Utility Vehicle 4WD</th>\n",
       "      <th>VClass_Subcompact Cars</th>\n",
       "      <th>VClass_Two Seaters</th>\n",
       "      <th>VClass_Vans, Cargo Type</th>\n",
       "      <th>VClass_Vans, Passenger Type</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>18.311667</td>\n",
       "      <td>16.0</td>\n",
       "      <td>18.0</td>\n",
       "      <td>3200.0</td>\n",
       "      <td>22.0</td>\n",
       "      <td>2004</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>15.695714</td>\n",
       "      <td>18.0</td>\n",
       "      <td>21.0</td>\n",
       "      <td>2750.0</td>\n",
       "      <td>26.0</td>\n",
       "      <td>2004</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>15.695714</td>\n",
       "      <td>18.0</td>\n",
       "      <td>21.0</td>\n",
       "      <td>2750.0</td>\n",
       "      <td>26.0</td>\n",
       "      <td>2004</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>15.695714</td>\n",
       "      <td>19.0</td>\n",
       "      <td>21.0</td>\n",
       "      <td>2750.0</td>\n",
       "      <td>25.0</td>\n",
       "      <td>2004</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>15.695714</td>\n",
       "      <td>18.0</td>\n",
       "      <td>21.0</td>\n",
       "      <td>2750.0</td>\n",
       "      <td>26.0</td>\n",
       "      <td>2004</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>...</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 35 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "   barrels08  city08  comb08  fuelCost08  highway08  year  score  \\\n",
       "0  18.311667    16.0    18.0      3200.0       22.0  2004      1   \n",
       "1  15.695714    18.0    21.0      2750.0       26.0  2004      1   \n",
       "2  15.695714    18.0    21.0      2750.0       26.0  2004      1   \n",
       "3  15.695714    19.0    21.0      2750.0       25.0  2004      1   \n",
       "4  15.695714    18.0    21.0      2750.0       26.0  2004      1   \n",
       "\n",
       "   fuelType1_Diesel  fuelType1_Green Fuel  fuelType1_Premium Gasoline  ...  \\\n",
       "0                 0                     0                           1  ...   \n",
       "1                 0                     0                           1  ...   \n",
       "2                 0                     0                           1  ...   \n",
       "3                 0                     0                           1  ...   \n",
       "4                 0                     0                           1  ...   \n",
       "\n",
       "   VClass_Sport Utility Vehicle - 2WD  VClass_Sport Utility Vehicle - 4WD  \\\n",
       "0                                   0                                   0   \n",
       "1                                   0                                   0   \n",
       "2                                   0                                   0   \n",
       "3                                   0                                   0   \n",
       "4                                   0                                   0   \n",
       "\n",
       "   VClass_Standard Pickup Trucks 2WD  VClass_Standard Pickup Trucks 4WD  \\\n",
       "0                                  0                                  0   \n",
       "1                                  0                                  0   \n",
       "2                                  0                                  0   \n",
       "3                                  0                                  0   \n",
       "4                                  0                                  0   \n",
       "\n",
       "   VClass_Standard Sport Utility Vehicle 2WD  \\\n",
       "0                                          0   \n",
       "1                                          0   \n",
       "2                                          0   \n",
       "3                                          0   \n",
       "4                                          0   \n",
       "\n",
       "   VClass_Standard Sport Utility Vehicle 4WD  VClass_Subcompact Cars  \\\n",
       "0                                          0                       0   \n",
       "1                                          0                       0   \n",
       "2                                          0                       0   \n",
       "3                                          0                       0   \n",
       "4                                          0                       0   \n",
       "\n",
       "   VClass_Two Seaters  VClass_Vans, Cargo Type  VClass_Vans, Passenger Type  \n",
       "0                   1                        0                            0  \n",
       "1                   1                        0                            0  \n",
       "2                   1                        0                            0  \n",
       "3                   1                        0                            0  \n",
       "4                   1                        0                            0  \n",
       "\n",
       "[5 rows x 35 columns]"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "ve_df.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "9411d858",
   "metadata": {},
   "source": [
    "## 4.2 Data splitting"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "b13e430d",
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.model_selection import train_test_split"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "e099be52",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "5     0.399005\n",
       "6     0.292548\n",
       "1     0.068217\n",
       "3     0.066046\n",
       "7     0.061493\n",
       "9     0.043213\n",
       "2     0.041357\n",
       "8     0.019085\n",
       "10    0.008825\n",
       "4     0.000210\n",
       "Name: score, dtype: float64"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y = ve_df['score'].copy()\n",
    "X = ve_df.drop(columns='score')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "b2465b46",
   "metadata": {},
   "outputs": [],
   "source": [
    "random_state = 0"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "b898ad0b",
   "metadata": {},
   "outputs": [],
   "source": [
    "X_train, X_test, y_train, y_test = train_test_split(\n",
    "    X, y, \n",
    "    test_size=0.3,\n",
    "    random_state=random_state,\n",
    "    stratify=y\n",
    ")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "4bf021fc",
   "metadata": {},
   "source": [
    "## 4.3 Training and Test Data Normalization"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "95f9f5f7",
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.preprocessing import MinMaxScaler"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "a8b1c9d9",
   "metadata": {},
   "outputs": [],
   "source": [
    "scaler = MinMaxScaler()\n",
    "X_train_norm = scaler.fit_transform(X_train.to_numpy())\n",
    "X_test_norm = scaler.transform(X_test.to_numpy())"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "cb6e7664",
   "metadata": {},
   "source": [
    "## 4.4 Model Tuning on Local System"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "31ac6348",
   "metadata": {},
   "source": [
    "### 4.4.1 Random Forests"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "45a4cb8c",
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.ensemble import RandomForestClassifier\n",
    "from sklearn.model_selection import GridSearchCV, RepeatedKFold"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "d08ee27f",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'f1_macro': 'f1_macro',\n",
       " 'precision_macro': 'precision_macro',\n",
       " 'recall_macro': 'recall_macro',\n",
       " 'balanced_accuracy': 'balanced_accuracy',\n",
       " 'kappa': make_scorer(cohen_kappa_score)}"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from sklearn.metrics import make_scorer, cohen_kappa_score\n",
    "\n",
    "score_avg = 'macro'\n",
    "kappa_scorer = make_scorer(cohen_kappa_score)\n",
    "scoring_metrics = {\n",
    "    'f1_' + score_avg: 'f1_' + score_avg, \n",
    "    'precision_' + score_avg: 'precision_' + score_avg, \n",
    "    'recall_' + score_avg: 'recall_' + score_avg,\n",
    "    'balanced_accuracy': 'balanced_accuracy',\n",
    "    'kappa': kappa_scorer\n",
    "}\n",
    "scoring_metrics"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 361,
   "id": "a474f7bd",
   "metadata": {},
   "outputs": [],
   "source": [
    "param_dict = {\n",
    "    'n_estimators': [500],\n",
    "    'max_features': [5, 8],\n",
    "    'max_depth': [None],\n",
    "    'max_samples': [0.8],\n",
    "    'min_samples_leaf': [1],\n",
    "    'criterion': ['gini', 'entropy'],\n",
    "    'class_weight': ['balanced'] #, 'balanced_subsample']\n",
    "}\n",
    "cv = RepeatedKFold(n_splits=5, n_repeats=1, random_state=random_state)\n",
    "model = RandomForestClassifier(n_jobs=-1, random_state=random_state)\n",
    "model_grid = GridSearchCV(\n",
    "    estimator=model, \n",
    "    param_grid=param_dict, \n",
    "    scoring=scoring_metrics, \n",
    "    n_jobs=-1, cv=cv, refit=scoring_metrics['f1_' + score_avg], \n",
    "    return_train_score=True\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 362,
   "id": "4b514b59",
   "metadata": {},
   "outputs": [],
   "source": [
    "search_rf = model_grid.fit(X_train_norm, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "162d18f1",
   "metadata": {},
   "outputs": [],
   "source": [
    "# User defined function for computing model metrics\n",
    "\n",
    "import numpy as np\n",
    "from sklearn.metrics import roc_auc_score, balanced_accuracy_score, f1_score, precision_score, recall_score, cohen_kappa_score\n",
    "from sklearn.metrics import classification_report\n",
    "\n",
    "def calc_error_metrics(gs_obj, X_test, y_test, score_avg='macro'):\n",
    "    \"\"\"\n",
    "    Calculate train, validation, test error metrics, and classification report (for the test data)\n",
    "    :param gs_obj: Fitted GridSearchCV object\n",
    "    :param X_test: X_test\n",
    "    :param y_test: y_test\n",
    "    :param score_avg: Scikit-learn scorer avergaging for multi-class classification\n",
    "    :return Pandas dataframe containing the error metrics and classification report\n",
    "    \"\"\"\n",
    "    \n",
    "    # auc_list = []\n",
    "    f1_list = []\n",
    "    precision_list = []\n",
    "    recall_list = []\n",
    "    ba_list = []\n",
    "    kappa_list = []\n",
    "    search_data = ['mean_train', 'mean_test']\n",
    "    scores = gs_obj.cv_results_\n",
    "    for sd in search_data:\n",
    "        # auc = scores['{}_roc_auc_{}'.format(sd, score_avg)].mean()\n",
    "        f1 = scores['{}_f1_{}'.format(sd, score_avg)].mean()\n",
    "        precision = scores['{}_precision_{}'.format(sd, score_avg)].mean()\n",
    "        recall = scores['{}_recall_{}'.format(sd, score_avg)].mean()\n",
    "        ba = scores['{}_balanced_accuracy'.format(sd)].mean()\n",
    "        kappa = scores['{}_kappa'.format(sd)].mean()\n",
    "        # auc_list.append(auc)\n",
    "        f1_list.append(f1)\n",
    "        precision_list.append(precision)\n",
    "        recall_list.append(recall)\n",
    "        ba_list.append(ba)\n",
    "        kappa_list.append(kappa)\n",
    "    best_model = gs_obj.best_estimator_\n",
    "    y_pred = best_model.predict(X_test)\n",
    "    # y_pred_proba = best_model.predict_proba(X_test)\n",
    "    labels = np.unique(y_pred)\n",
    "    # test_auc = roc_auc_score(y_test, y_pred, labels=labels, average=score_avg)\n",
    "    test_f1 = f1_score(y_test, y_pred, labels=labels, average=score_avg)\n",
    "    test_precision = precision_score(y_test, y_pred, labels=labels, average=score_avg)\n",
    "    test_recall = recall_score(y_test, y_pred, labels=labels, average=score_avg)\n",
    "    test_ba = balanced_accuracy_score(y_test, y_pred)\n",
    "    test_kappa = cohen_kappa_score(y_test, y_pred, labels=labels)\n",
    "    # auc_list.append(test_auc)\n",
    "    f1_list.append(test_f1)\n",
    "    precision_list.append(test_precision)\n",
    "    recall_list.append(test_recall)\n",
    "    ba_list.append(test_ba)\n",
    "    kappa_list.append(test_kappa)\n",
    "    results_dict = {\n",
    "        'Data': ['Train', 'Validation', 'Test'],\n",
    "        # 'ROC_AUC': auc_list,\n",
    "        'F1 Score': f1_list,\n",
    "        'Precision': precision_list,\n",
    "        'Recall': recall_list,\n",
    "        'Balanced Accuracy': ba_list,\n",
    "        'Kappa Score': kappa_list\n",
    "    }\n",
    "    results_df = pd.DataFrame(data=results_dict)\n",
    "    report = classification_report(y_test, y_pred, target_names=labels.astype(str), labels=labels)\n",
    "    return results_df, report    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 364,
   "id": "f03004eb",
   "metadata": {},
   "outputs": [],
   "source": [
    "metrics_df, report = calc_error_metrics(search_rf, X_test_norm, y_test, score_avg)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 365,
   "id": "7b228389",
   "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>Data</th>\n",
       "      <th>F1 Score</th>\n",
       "      <th>Precision</th>\n",
       "      <th>Recall</th>\n",
       "      <th>Balanced Accuracy</th>\n",
       "      <th>Kappa Score</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>Train</td>\n",
       "      <td>0.777282</td>\n",
       "      <td>0.726165</td>\n",
       "      <td>0.869580</td>\n",
       "      <td>0.869580</td>\n",
       "      <td>0.671342</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>Validation</td>\n",
       "      <td>0.467633</td>\n",
       "      <td>0.449835</td>\n",
       "      <td>0.499578</td>\n",
       "      <td>0.509674</td>\n",
       "      <td>0.255996</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>Test</td>\n",
       "      <td>0.478083</td>\n",
       "      <td>0.447769</td>\n",
       "      <td>0.532153</td>\n",
       "      <td>0.532153</td>\n",
       "      <td>0.229528</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "         Data  F1 Score  Precision    Recall  Balanced Accuracy  Kappa Score\n",
       "0       Train  0.777282   0.726165  0.869580           0.869580     0.671342\n",
       "1  Validation  0.467633   0.449835  0.499578           0.509674     0.255996\n",
       "2        Test  0.478083   0.447769  0.532153           0.532153     0.229528"
      ]
     },
     "execution_count": 365,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "metrics_df"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 366,
   "id": "cac76d9a",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "           1       0.56      0.61      0.59       584\n",
      "           2       0.40      0.50      0.44       354\n",
      "           3       0.65      0.68      0.67       566\n",
      "           4       0.50      1.00      0.67         2\n",
      "           5       0.47      0.40      0.43      3418\n",
      "           6       0.36      0.34      0.35      2506\n",
      "           7       0.35      0.42      0.38       527\n",
      "           8       0.12      0.23      0.16       164\n",
      "           9       0.09      0.14      0.11       370\n",
      "          10       0.97      1.00      0.99        76\n",
      "\n",
      "    accuracy                           0.41      8567\n",
      "   macro avg       0.45      0.53      0.48      8567\n",
      "weighted avg       0.43      0.41      0.42      8567\n",
      "\n"
     ]
    }
   ],
   "source": [
    "print(report)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 302,
   "id": "502c39b1",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{'class_weight': 'balanced', 'criterion': 'entropy', 'max_depth': None, 'max_features': 5, 'max_samples': None, 'min_samples_leaf': 1, 'n_estimators': 500}\n"
     ]
    }
   ],
   "source": [
    "print(search_rf.best_params_)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f70d94f4",
   "metadata": {},
   "source": [
    "## 4.5 Model Tuning on the Missouri S&T Foundry HPC"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "7d280d35",
   "metadata": {},
   "source": [
    "We use Dask to perform model tuning on a distributed computing environment"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "565538e5",
   "metadata": {},
   "source": [
    "### 4.5.1 LightGBM"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "f6810262",
   "metadata": {},
   "outputs": [],
   "source": [
    "from dask_ml.model_selection import GridSearchCV as DaskGCV\n",
    "from dask.distributed import Client\n",
    "from dask_jobqueue import SLURMCluster\n",
    "from lightgbm import LGBMClassifier"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "id": "e559aa05",
   "metadata": {},
   "outputs": [],
   "source": [
    "param_dict = {\n",
    "    'boosting_type': ['gbdt'],\n",
    "    'n_estimators': [300, 500],\n",
    "    'max_depth': [10, 20, 30, -1],\n",
    "    'max_bin': [127, 255],\n",
    "    'min_data_in_bin': [3, 10, 20],\n",
    "    'learning_rate': [0.05, 0.1],\n",
    "    'subsample': [1, 0.8],\n",
    "    'colsample_bytree': [1, 0.5],\n",
    "    'colsample_bynode': [1],\n",
    "    'path_smooth': [0, 0.1, 0.2],\n",
    "    'num_leaves': [31, 63],\n",
    "    'min_child_samples': [20],\n",
    "}\n",
    "cv = RepeatedKFold(n_splits=5, n_repeats=1, random_state=random_state)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "id": "a14d5e1c",
   "metadata": {},
   "outputs": [],
   "source": [
    "model = LGBMClassifier(\n",
    "    tree_learner='feature', \n",
    "    class_weight='balanced',\n",
    "    random_state=random_state,\n",
    "    # deterministic=True, force_row_wise=True\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "id": "9f65b210",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Waiting for dask workers...\n"
     ]
    }
   ],
   "source": [
    "cluster = SLURMCluster(\n",
    "    cores=40,\n",
    "    processes=1,\n",
    "    memory=\"50G\",\n",
    "    walltime=\"02:00:00\",\n",
    "    interface='ib0',\n",
    "    env_extra=['#SBATCH --out=IST-5520-Foundry-Dask-%j.out']\n",
    ")\n",
    "cluster.adapt(\n",
    "    minimum=10, maximum=50,\n",
    "    minimum_jobs=10, maximum_jobs=50,\n",
    "    minimum_memory='40G', maximum_memory='50G'\n",
    ")\n",
    "dask_client = Client(cluster)\n",
    "print('Waiting for dask workers...')\n",
    "dask_client.wait_for_workers(10)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "id": "bf71c9dc",
   "metadata": {},
   "outputs": [],
   "source": [
    "model_grid = DaskGCV(\n",
    "    estimator=model, \n",
    "    param_grid=param_dict, \n",
    "    scoring=scoring_metrics, \n",
    "    n_jobs=-1, cv=cv, \n",
    "    refit=scoring_metrics['f1_' + score_avg], \n",
    "    return_train_score=True\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "08d1e5ff",
   "metadata": {},
   "outputs": [],
   "source": [
    "search_dask_lgbm = model_grid.fit(X_train_norm, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "id": "31c29ef0",
   "metadata": {},
   "outputs": [],
   "source": [
    "metrics_df, report = calc_error_metrics(search_dask_lgbm, X_test_norm, y_test, score_avg)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "id": "a2f2208d",
   "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>Data</th>\n",
       "      <th>F1 Score</th>\n",
       "      <th>Precision</th>\n",
       "      <th>Recall</th>\n",
       "      <th>Balanced Accuracy</th>\n",
       "      <th>Kappa Score</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>Train</td>\n",
       "      <td>0.702728</td>\n",
       "      <td>0.644762</td>\n",
       "      <td>0.839862</td>\n",
       "      <td>0.839862</td>\n",
       "      <td>0.540943</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>Validation</td>\n",
       "      <td>0.481861</td>\n",
       "      <td>0.452456</td>\n",
       "      <td>0.556968</td>\n",
       "      <td>0.568588</td>\n",
       "      <td>0.324939</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>Test</td>\n",
       "      <td>0.514768</td>\n",
       "      <td>0.471515</td>\n",
       "      <td>0.633385</td>\n",
       "      <td>0.633385</td>\n",
       "      <td>0.329062</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "         Data  F1 Score  Precision    Recall  Balanced Accuracy  Kappa Score\n",
       "0       Train  0.702728   0.644762  0.839862           0.839862     0.540943\n",
       "1  Validation  0.481861   0.452456  0.556968           0.568588     0.324939\n",
       "2        Test  0.514768   0.471515  0.633385           0.633385     0.329062"
      ]
     },
     "execution_count": 22,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "metrics_df"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "id": "94357e53",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "           1       0.50      0.66      0.57       584\n",
      "           2       0.36      0.67      0.47       354\n",
      "           3       0.55      0.81      0.65       566\n",
      "           4       0.50      1.00      0.67         2\n",
      "           5       0.64      0.30      0.41      3418\n",
      "           6       0.49      0.47      0.48      2506\n",
      "           7       0.39      0.54      0.45       527\n",
      "           8       0.15      0.48      0.23       164\n",
      "           9       0.15      0.41      0.22       370\n",
      "          10       0.97      1.00      0.99        76\n",
      "\n",
      "    accuracy                           0.45      8567\n",
      "   macro avg       0.47      0.63      0.51      8567\n",
      "weighted avg       0.53      0.45      0.46      8567\n",
      "\n"
     ]
    }
   ],
   "source": [
    "print(report)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "id": "d5476307",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{'boosting_type': 'gbdt', 'colsample_bynode': 1, 'colsample_bytree': 1, 'learning_rate': 0.05, 'max_bin': 127, 'max_depth': 10, 'min_child_samples': 20, 'min_data_in_bin': 3, 'n_estimators': 300, 'num_leaves': 31, 'path_smooth': 0.2, 'subsample': 1}\n"
     ]
    }
   ],
   "source": [
    "print(search_dask_lgbm.best_params_)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 240,
   "id": "54410732",
   "metadata": {},
   "outputs": [],
   "source": [
    "dask_client.shutdown()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "988ac8b7",
   "metadata": {},
   "source": [
    "### 4.5.2 Random Forests "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 232,
   "id": "3df2aa6b",
   "metadata": {},
   "outputs": [],
   "source": [
    "param_dict = {\n",
    "    'n_estimators': [300, 500],\n",
    "    'max_features': [3, 5, 10, 20],\n",
    "    'max_depth': [6, 10, None],\n",
    "    'max_samples': [0.8, 0.5, 0.9, None],\n",
    "    'min_samples_leaf': [1],\n",
    "    'criterion': ['gini', 'entropy'],\n",
    "    'class_weight': ['balanced', 'balanced_subsample']\n",
    "}\n",
    "cv = RepeatedKFold(n_splits=5, n_repeats=1, random_state=random_state)\n",
    "model = RandomForestClassifier(\n",
    "    n_jobs=-1, \n",
    "    random_state=random_state)\n",
    "model_grid = DaskGCV(\n",
    "    estimator=model, \n",
    "    param_grid=param_dict, \n",
    "    scoring=scoring_metrics, \n",
    "    n_jobs=-1, cv=cv, refit=scoring_metrics['f1_' + score_avg], \n",
    "    return_train_score=True\n",
    ")\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 233,
   "id": "6fde6a48",
   "metadata": {},
   "outputs": [],
   "source": [
    "model_grid = DaskGCV(\n",
    "    estimator=model, \n",
    "    param_grid=param_dict, \n",
    "    scoring=scoring_metrics, \n",
    "    n_jobs=-1, cv=cv, \n",
    "    refit=scoring_metrics['f1_' + score_avg], \n",
    "    return_train_score=True\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 234,
   "id": "e8c8b457",
   "metadata": {},
   "outputs": [],
   "source": [
    "search_rf = model_grid.fit(X_train_norm, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 235,
   "id": "f66e6016",
   "metadata": {},
   "outputs": [],
   "source": [
    "metrics_df, report = calc_error_metrics(search_rf, X_test_norm, y_test, score_avg)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 236,
   "id": "ea7dbc5d",
   "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>Data</th>\n",
       "      <th>F1 Score</th>\n",
       "      <th>Precision</th>\n",
       "      <th>Recall</th>\n",
       "      <th>Balanced Accuracy</th>\n",
       "      <th>Kappa Score</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>Train</td>\n",
       "      <td>0.578722</td>\n",
       "      <td>0.555383</td>\n",
       "      <td>0.714078</td>\n",
       "      <td>0.714078</td>\n",
       "      <td>0.422408</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>Validation</td>\n",
       "      <td>0.436231</td>\n",
       "      <td>0.426816</td>\n",
       "      <td>0.533302</td>\n",
       "      <td>0.547823</td>\n",
       "      <td>0.263565</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>Test</td>\n",
       "      <td>0.408527</td>\n",
       "      <td>0.398950</td>\n",
       "      <td>0.420700</td>\n",
       "      <td>0.420700</td>\n",
       "      <td>0.227434</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "         Data  F1 Score  Precision    Recall  Balanced Accuracy  Kappa Score\n",
       "0       Train  0.578722   0.555383  0.714078           0.714078     0.422408\n",
       "1  Validation  0.436231   0.426816  0.533302           0.547823     0.263565\n",
       "2        Test  0.408527   0.398950  0.420700           0.420700     0.227434"
      ]
     },
     "execution_count": 236,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "metrics_df"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 237,
   "id": "205fbf49",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "           1       0.57      0.61      0.59       584\n",
      "           2       0.40      0.45      0.43       354\n",
      "           3       0.66      0.69      0.67       566\n",
      "           4       0.00      0.00      0.00         2\n",
      "           5       0.47      0.43      0.45      3418\n",
      "           6       0.35      0.35      0.35      2506\n",
      "           7       0.37      0.39      0.38       527\n",
      "           8       0.12      0.19      0.14       164\n",
      "           9       0.08      0.09      0.08       370\n",
      "          10       0.97      1.00      0.99        76\n",
      "\n",
      "    accuracy                           0.42      8567\n",
      "   macro avg       0.40      0.42      0.41      8567\n",
      "weighted avg       0.43      0.42      0.42      8567\n",
      "\n"
     ]
    }
   ],
   "source": [
    "print(report)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 238,
   "id": "63b3a0be",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{'class_weight': 'balanced_subsample', 'criterion': 'entropy', 'max_depth': None, 'max_features': 10, 'max_samples': 0.5, 'min_samples_leaf': 1, 'n_estimators': 500}\n"
     ]
    }
   ],
   "source": [
    "print(search_rf.best_params_)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "45ef0ec7",
   "metadata": {},
   "source": [
    "### 4.5.3 Support Vector Machine"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "id": "a967473a",
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.svm import SVC"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "id": "53000d96",
   "metadata": {},
   "outputs": [],
   "source": [
    "param_dict = {\n",
    "    'C': [0.5, 0.8, 1],\n",
    "    'gamma': ['scale', 'auto', 0.8],\n",
    "    'tol': [1e-3, 1e-2],\n",
    "    'break_ties': [True, False]\n",
    "}\n",
    "cv = RepeatedKFold(n_splits=5, n_repeats=1, random_state=random_state)\n",
    "model = SVC(class_weight='balanced', random_state=random_state)\n",
    "model_grid = DaskGCV(\n",
    "    estimator=model, \n",
    "    param_grid=param_dict, \n",
    "    scoring=scoring_metrics, \n",
    "    n_jobs=-1, cv=cv, refit=scoring_metrics['f1_' + score_avg], \n",
    "    return_train_score=True\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "id": "76d40a25",
   "metadata": {},
   "outputs": [],
   "source": [
    "search_svm = model_grid.fit(X_train_norm, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "id": "7807efa4",
   "metadata": {},
   "outputs": [],
   "source": [
    "metrics_df, report = calc_error_metrics(search_svm, X_test_norm, y_test, score_avg)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "id": "2ce083f1",
   "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>Data</th>\n",
       "      <th>F1 Score</th>\n",
       "      <th>Precision</th>\n",
       "      <th>Recall</th>\n",
       "      <th>Balanced Accuracy</th>\n",
       "      <th>Kappa Score</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>Train</td>\n",
       "      <td>0.370993</td>\n",
       "      <td>0.365296</td>\n",
       "      <td>0.577352</td>\n",
       "      <td>0.577352</td>\n",
       "      <td>0.231768</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>Validation</td>\n",
       "      <td>0.356233</td>\n",
       "      <td>0.355257</td>\n",
       "      <td>0.511651</td>\n",
       "      <td>0.532175</td>\n",
       "      <td>0.220769</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>Test</td>\n",
       "      <td>0.399593</td>\n",
       "      <td>0.386629</td>\n",
       "      <td>0.578398</td>\n",
       "      <td>0.578398</td>\n",
       "      <td>0.250525</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "         Data  F1 Score  Precision    Recall  Balanced Accuracy  Kappa Score\n",
       "0       Train  0.370993   0.365296  0.577352           0.577352     0.231768\n",
       "1  Validation  0.356233   0.355257  0.511651           0.532175     0.220769\n",
       "2        Test  0.399593   0.386629  0.578398           0.578398     0.250525"
      ]
     },
     "execution_count": 20,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "metrics_df"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "id": "97e47296",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "           1       0.46      0.46      0.46       584\n",
      "           2       0.24      0.75      0.36       354\n",
      "           3       0.41      0.71      0.52       566\n",
      "           4       0.13      1.00      0.24         2\n",
      "           5       0.63      0.23      0.34      3418\n",
      "           6       0.50      0.30      0.38      2506\n",
      "           7       0.33      0.54      0.41       527\n",
      "           8       0.06      0.37      0.10       164\n",
      "           9       0.14      0.43      0.21       370\n",
      "          10       0.97      1.00      0.99        76\n",
      "\n",
      "    accuracy                           0.36      8567\n",
      "   macro avg       0.39      0.58      0.40      8567\n",
      "weighted avg       0.50      0.36      0.37      8567\n",
      "\n"
     ]
    }
   ],
   "source": [
    "print(report)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "id": "bfc57279",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{'C': 1, 'break_ties': True, 'gamma': 0.8, 'tol': 0.01}\n"
     ]
    }
   ],
   "source": [
    "print(search_svm.best_params_)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "8b40b24b",
   "metadata": {},
   "source": [
    "### 4.5.4 Extra Trees Classifier"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "id": "2d5adeaa",
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.ensemble import ExtraTreesClassifier"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "id": "993272a4",
   "metadata": {},
   "outputs": [],
   "source": [
    "param_dict = {\n",
    "    'n_estimators': [300, 500],\n",
    "    'max_features': [3, 5, 7, 10],\n",
    "    'max_depth': [6, 10, None],\n",
    "    'max_samples': [0.8, 0.5, 0.9, None],\n",
    "    'min_samples_leaf': [1],\n",
    "    'criterion': ['gini', 'entropy'],\n",
    "    'class_weight': ['balanced', 'balanced_subsample']\n",
    "}\n",
    "cv = RepeatedKFold(n_splits=5, n_repeats=1, random_state=random_state)\n",
    "model = ExtraTreesClassifier(\n",
    "    class_weight='balanced', \n",
    "    random_state=random_state,\n",
    "    n_jobs=-1\n",
    ")\n",
    "model_grid = DaskGCV(\n",
    "    estimator=model, \n",
    "    param_grid=param_dict, \n",
    "    scoring=scoring_metrics, \n",
    "    n_jobs=-1, cv=cv, refit=scoring_metrics['f1_' + score_avg], \n",
    "    return_train_score=True\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "id": "f5b1e335",
   "metadata": {},
   "outputs": [],
   "source": [
    "search_et = model_grid.fit(X_train_norm, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "id": "c38700cf",
   "metadata": {},
   "outputs": [],
   "source": [
    "metrics_df, report = calc_error_metrics(search_et, X_test_norm, y_test, score_avg)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "id": "60adbb49",
   "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>Data</th>\n",
       "      <th>F1 Score</th>\n",
       "      <th>Precision</th>\n",
       "      <th>Recall</th>\n",
       "      <th>Balanced Accuracy</th>\n",
       "      <th>Kappa Score</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>Train</td>\n",
       "      <td>0.533500</td>\n",
       "      <td>0.514130</td>\n",
       "      <td>0.709851</td>\n",
       "      <td>0.709851</td>\n",
       "      <td>0.401488</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>Validation</td>\n",
       "      <td>0.400826</td>\n",
       "      <td>0.395748</td>\n",
       "      <td>0.519259</td>\n",
       "      <td>0.537028</td>\n",
       "      <td>0.239507</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>Test</td>\n",
       "      <td>0.462869</td>\n",
       "      <td>0.428880</td>\n",
       "      <td>0.533174</td>\n",
       "      <td>0.533174</td>\n",
       "      <td>0.221032</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "         Data  F1 Score  Precision    Recall  Balanced Accuracy  Kappa Score\n",
       "0       Train  0.533500   0.514130  0.709851           0.709851     0.401488\n",
       "1  Validation  0.400826   0.395748  0.519259           0.537028     0.239507\n",
       "2        Test  0.462869   0.428880  0.533174           0.533174     0.221032"
      ]
     },
     "execution_count": 33,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "metrics_df"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "id": "c5b2ecf6",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "           1       0.55      0.61      0.58       584\n",
      "           2       0.39      0.53      0.45       354\n",
      "           3       0.62      0.67      0.64       566\n",
      "           4       0.40      1.00      0.57         2\n",
      "           5       0.46      0.36      0.40      3418\n",
      "           6       0.36      0.36      0.36      2506\n",
      "           7       0.34      0.40      0.37       527\n",
      "           8       0.12      0.24      0.16       164\n",
      "           9       0.10      0.16      0.12       370\n",
      "          10       0.96      1.00      0.98        76\n",
      "\n",
      "    accuracy                           0.40      8567\n",
      "   macro avg       0.43      0.53      0.46      8567\n",
      "weighted avg       0.42      0.40      0.40      8567\n",
      "\n"
     ]
    }
   ],
   "source": [
    "print(report)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "id": "40c12217",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{'class_weight': 'balanced', 'criterion': 'entropy', 'max_depth': None, 'max_features': 10, 'max_samples': 0.8, 'min_samples_leaf': 1, 'n_estimators': 300}\n"
     ]
    }
   ],
   "source": [
    "print(search_et.best_params_)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "4d53cb9c",
   "metadata": {},
   "source": [
    "### 4.5.5 MLP Classifier"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "id": "ad6e26e3",
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.neural_network import MLPClassifier"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "id": "dcf50454",
   "metadata": {},
   "outputs": [],
   "source": [
    "param_dict = {\n",
    "    'hidden_layer_sizes': [(512, 256), (256, 128, 64), (64, 32, 16, 8)],\n",
    "    'learning_rate_init': [1e-3, 1e-2],\n",
    "    'max_iter': [100, 200],\n",
    "}\n",
    "cv = RepeatedKFold(n_splits=5, n_repeats=1, random_state=random_state)\n",
    "model = MLPClassifier(random_state=random_state)\n",
    "model_grid = DaskGCV(\n",
    "    estimator=model, \n",
    "    param_grid=param_dict, \n",
    "    scoring=scoring_metrics, \n",
    "    n_jobs=-1, cv=cv, refit=scoring_metrics['f1_' + score_avg], \n",
    "    return_train_score=True\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "id": "6e7e1201",
   "metadata": {},
   "outputs": [],
   "source": [
    "search_mlp = model_grid.fit(X_train_norm, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "id": "b1e5bbdd",
   "metadata": {},
   "outputs": [],
   "source": [
    "metrics_df, report = calc_error_metrics(search_mlp, X_test_norm, y_test, score_avg)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "id": "d9d5b67a",
   "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>Data</th>\n",
       "      <th>F1 Score</th>\n",
       "      <th>Precision</th>\n",
       "      <th>Recall</th>\n",
       "      <th>Balanced Accuracy</th>\n",
       "      <th>Kappa Score</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>Train</td>\n",
       "      <td>0.526306</td>\n",
       "      <td>0.590153</td>\n",
       "      <td>0.513441</td>\n",
       "      <td>0.513441</td>\n",
       "      <td>0.385971</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>Validation</td>\n",
       "      <td>0.450087</td>\n",
       "      <td>0.489226</td>\n",
       "      <td>0.445315</td>\n",
       "      <td>0.453457</td>\n",
       "      <td>0.324013</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>Test</td>\n",
       "      <td>0.483473</td>\n",
       "      <td>0.530263</td>\n",
       "      <td>0.476535</td>\n",
       "      <td>0.428881</td>\n",
       "      <td>0.326520</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "         Data  F1 Score  Precision    Recall  Balanced Accuracy  Kappa Score\n",
       "0       Train  0.526306   0.590153  0.513441           0.513441     0.385971\n",
       "1  Validation  0.450087   0.489226  0.445315           0.453457     0.324013\n",
       "2        Test  0.483473   0.530263  0.476535           0.428881     0.326520"
      ]
     },
     "execution_count": 21,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "metrics_df"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "id": "e195b522",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "           1       0.57      0.59      0.58       584\n",
      "           2       0.49      0.38      0.43       354\n",
      "           3       0.57      0.72      0.64       566\n",
      "           5       0.52      0.67      0.59      3418\n",
      "           6       0.48      0.38      0.43      2506\n",
      "           7       0.53      0.37      0.43       527\n",
      "           8       0.39      0.09      0.15       164\n",
      "           9       0.25      0.09      0.13       370\n",
      "          10       0.97      1.00      0.99        76\n",
      "\n",
      "   micro avg       0.52      0.52      0.52      8565\n",
      "   macro avg       0.53      0.48      0.48      8565\n",
      "weighted avg       0.51      0.52      0.50      8565\n",
      "\n"
     ]
    }
   ],
   "source": [
    "print(report)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "id": "0ea4958b",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{'hidden_layer_sizes': (512, 256), 'learning_rate_init': 0.001, 'max_iter': 100}\n"
     ]
    }
   ],
   "source": [
    "print(search_mlp.best_params_)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "id": "bbaf21b4",
   "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": [
    "import matplotlib.pyplot as plt\n",
    "plt.plot(search_mlp.best_estimator_.loss_curve_)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "b211f062",
   "metadata": {},
   "source": [
    "#### 4.5.6 Boosted Random Forests"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "id": "bd14399d",
   "metadata": {},
   "outputs": [],
   "source": [
    "param_dict = {\n",
    "    'n_estimators': [300, 500],\n",
    "    'max_depth': [10, 20, -1],\n",
    "    'max_bin': [127, 255],\n",
    "    'min_data_in_bin': [3, 10, 20],\n",
    "    'learning_rate': [0.05, 0.1],\n",
    "    'subsample': [0.9, 0.5],\n",
    "    'colsample_bytree': [1, 0.5],\n",
    "    'colsample_bynode': [1],\n",
    "    'path_smooth': [0, 0.1, 0.2],\n",
    "    'num_leaves': [31, 63],\n",
    "    'min_child_samples': [20],\n",
    "}\n",
    "cv = RepeatedKFold(n_splits=5, n_repeats=1, random_state=random_state)\n",
    "model = LGBMClassifier(\n",
    "    boosting_type='rf',\n",
    "    bagging_freq=1,\n",
    "    class_weight='balanced', \n",
    "    random_state=random_state\n",
    ")\n",
    "model_grid = DaskGCV(\n",
    "    estimator=model, \n",
    "    param_grid=param_dict, \n",
    "    scoring=scoring_metrics, \n",
    "    n_jobs=-1, cv=cv, refit=scoring_metrics['f1_' + score_avg], \n",
    "    return_train_score=True\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "id": "ba6e451e",
   "metadata": {},
   "outputs": [],
   "source": [
    "search_brf = model_grid.fit(X_train_norm, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "id": "e452697c",
   "metadata": {},
   "outputs": [],
   "source": [
    "metrics_df, report = calc_error_metrics(search_brf, X_test_norm, y_test, score_avg)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "id": "c324652b",
   "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>Data</th>\n",
       "      <th>F1 Score</th>\n",
       "      <th>Precision</th>\n",
       "      <th>Recall</th>\n",
       "      <th>Balanced Accuracy</th>\n",
       "      <th>Kappa Score</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>Train</td>\n",
       "      <td>0.457200</td>\n",
       "      <td>0.431045</td>\n",
       "      <td>0.675145</td>\n",
       "      <td>0.675145</td>\n",
       "      <td>0.331778</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>Validation</td>\n",
       "      <td>0.424028</td>\n",
       "      <td>0.404417</td>\n",
       "      <td>0.578920</td>\n",
       "      <td>0.602269</td>\n",
       "      <td>0.296177</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>Test</td>\n",
       "      <td>0.449043</td>\n",
       "      <td>0.417397</td>\n",
       "      <td>0.644578</td>\n",
       "      <td>0.644578</td>\n",
       "      <td>0.309786</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "         Data  F1 Score  Precision    Recall  Balanced Accuracy  Kappa Score\n",
       "0       Train  0.457200   0.431045  0.675145           0.675145     0.331778\n",
       "1  Validation  0.424028   0.404417  0.578920           0.602269     0.296177\n",
       "2        Test  0.449043   0.417397  0.644578           0.644578     0.309786"
      ]
     },
     "execution_count": 20,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "metrics_df"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "id": "266290cc",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "           1       0.46      0.62      0.53       584\n",
      "           2       0.33      0.72      0.45       354\n",
      "           3       0.49      0.80      0.60       566\n",
      "           4       0.10      1.00      0.17         2\n",
      "           5       0.64      0.26      0.37      3418\n",
      "           6       0.49      0.43      0.46      2506\n",
      "           7       0.38      0.50      0.43       527\n",
      "           8       0.14      0.66      0.23       164\n",
      "           9       0.18      0.46      0.26       370\n",
      "          10       0.97      1.00      0.99        76\n",
      "\n",
      "    accuracy                           0.43      8567\n",
      "   macro avg       0.42      0.64      0.45      8567\n",
      "weighted avg       0.52      0.43      0.43      8567\n",
      "\n"
     ]
    }
   ],
   "source": [
    "print(report)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "id": "b95b4452",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{'colsample_bynode': 1, 'colsample_bytree': 1, 'learning_rate': 0.05, 'max_bin': 255, 'max_depth': -1, 'min_child_samples': 20, 'min_data_in_bin': 3, 'n_estimators': 500, 'num_leaves': 63, 'path_smooth': 0, 'subsample': 0.9}\n"
     ]
    }
   ],
   "source": [
    "print(search_brf.best_params_)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "6975b25b",
   "metadata": {},
   "source": [
    "### 4.5.7 Multinomial Logistic Regression"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 212,
   "id": "9c17200f",
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.linear_model import LogisticRegression"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 213,
   "id": "5362deee",
   "metadata": {},
   "outputs": [],
   "source": [
    "y = ve_df['score'].copy()\n",
    "X = ve_df.drop(columns='score')\n",
    "X['const'] = 1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 214,
   "id": "1312441b",
   "metadata": {},
   "outputs": [],
   "source": [
    "X_train, X_test, y_train, y_test = train_test_split(\n",
    "    X, y, \n",
    "    test_size=0.3,\n",
    "    random_state=random_state,\n",
    "    stratify=y\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 215,
   "id": "4296c50d",
   "metadata": {},
   "outputs": [],
   "source": [
    "scaler = MinMaxScaler()\n",
    "X_train_norm = scaler.fit_transform(X_train.to_numpy())\n",
    "X_test_norm = scaler.transform(X_test.to_numpy())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 216,
   "id": "b956c6eb",
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.model_selection import RepeatedKFold\n",
    "param_dict = {\n",
    "    'tol': [1e-4, 1e-3, 1e-2],\n",
    "    'C': [0.5, 0.8, 1],\n",
    "    'solver': ['saga', 'lbfgs'],\n",
    "    'max_iter': [100, 200, 300, 500]\n",
    "}\n",
    "cv = RepeatedKFold(n_splits=5, n_repeats=1, random_state=random_state)\n",
    "model = LogisticRegression(\n",
    "    class_weight='balanced', \n",
    "    random_state=random_state,\n",
    "    multi_class='multinomial',\n",
    "    n_jobs=-1\n",
    ")\n",
    "model_grid = DaskGCV(\n",
    "    estimator=model, \n",
    "    param_grid=param_dict, \n",
    "    scoring=scoring_metrics, \n",
    "    n_jobs=-1, cv=cv, refit=scoring_metrics['f1_' + score_avg], \n",
    "    return_train_score=True\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 217,
   "id": "03826b7d",
   "metadata": {},
   "outputs": [],
   "source": [
    "search_mlr = model_grid.fit(X_train_norm, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 218,
   "id": "e31df96c",
   "metadata": {},
   "outputs": [],
   "source": [
    "metrics_df, report = calc_error_metrics(search_mlr, X_test_norm, y_test, score_avg)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 219,
   "id": "51f90bb4",
   "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>Data</th>\n",
       "      <th>F1 Score</th>\n",
       "      <th>Precision</th>\n",
       "      <th>Recall</th>\n",
       "      <th>Balanced Accuracy</th>\n",
       "      <th>Kappa Score</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>Train</td>\n",
       "      <td>0.331403</td>\n",
       "      <td>0.329090</td>\n",
       "      <td>0.509821</td>\n",
       "      <td>0.509821</td>\n",
       "      <td>0.189551</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>Validation</td>\n",
       "      <td>0.326589</td>\n",
       "      <td>0.326447</td>\n",
       "      <td>0.469016</td>\n",
       "      <td>0.483657</td>\n",
       "      <td>0.183054</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>Test</td>\n",
       "      <td>0.346034</td>\n",
       "      <td>0.353011</td>\n",
       "      <td>0.454485</td>\n",
       "      <td>0.409037</td>\n",
       "      <td>0.161838</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "         Data  F1 Score  Precision    Recall  Balanced Accuracy  Kappa Score\n",
       "0       Train  0.331403   0.329090  0.509821           0.509821     0.189551\n",
       "1  Validation  0.326589   0.326447  0.469016           0.483657     0.183054\n",
       "2        Test  0.346034   0.353011  0.454485           0.409037     0.161838"
      ]
     },
     "execution_count": 219,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "metrics_df"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 220,
   "id": "d1f82663",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "           1       0.13      0.31      0.18       584\n",
      "           2       0.25      0.61      0.35       354\n",
      "           3       0.32      0.53      0.40       566\n",
      "           5       0.49      0.19      0.27      3418\n",
      "           6       0.51      0.19      0.28      2506\n",
      "           7       0.32      0.35      0.33       527\n",
      "           8       0.06      0.49      0.11       164\n",
      "           9       0.14      0.42      0.21       370\n",
      "          10       0.96      1.00      0.98        76\n",
      "\n",
      "   micro avg       0.27      0.27      0.27      8565\n",
      "   macro avg       0.35      0.45      0.35      8565\n",
      "weighted avg       0.42      0.27      0.28      8565\n",
      "\n"
     ]
    }
   ],
   "source": [
    "print(report)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 221,
   "id": "3a115a43",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{'C': 1, 'max_iter': 300, 'solver': 'saga', 'tol': 0.0001}\n"
     ]
    }
   ],
   "source": [
    "print(search_mlr.best_params_)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "419b9a81",
   "metadata": {},
   "source": [
    "# 5. Final Model"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "0788592f",
   "metadata": {},
   "source": [
    "## 5.1 Get Training and Test Data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "6a7a9f5c",
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "59ffed38",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Note that we are not using dimension reduced data sets from PCA or correlation analysis.\n",
    "ve_df_all = pd.read_csv('data/Vehicle_Emissions.csv')\n",
    "ve_df = pd.read_csv('data/VE_Cleaned.csv')\n",
    "ve_df['VClass'] = ve_df_all['VClass'] # Getting original VClass"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "42bf4a46",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Creating dummies\n",
    "\n",
    "ve_df = pd.get_dummies(ve_df, columns=['fuelType1', 'VClass'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "09945108",
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.model_selection import train_test_split"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "b59acf5f",
   "metadata": {},
   "outputs": [],
   "source": [
    "y = ve_df['score'].copy()\n",
    "X = ve_df.drop(columns='score')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "3910cc3b",
   "metadata": {},
   "outputs": [],
   "source": [
    "random_state = 0"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "b46bd761",
   "metadata": {},
   "outputs": [],
   "source": [
    "X_train, X_test, y_train, y_test = train_test_split(\n",
    "    X, y, \n",
    "    test_size=0.3,\n",
    "    random_state=random_state,\n",
    "    stratify=y\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "1e3d6217",
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.preprocessing import MinMaxScaler"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "03747029",
   "metadata": {},
   "outputs": [],
   "source": [
    "scaler = MinMaxScaler()\n",
    "X_train_norm = scaler.fit_transform(X_train.to_numpy())\n",
    "X_test_norm = scaler.transform(X_test.to_numpy())"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "5cf0575f",
   "metadata": {},
   "source": [
    "## 5.1 Fitting with the best parameters"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "d5a6a6e4",
   "metadata": {},
   "outputs": [],
   "source": [
    "param_dict = {\n",
    "    'boosting_type': 'gbdt', \n",
    "    'colsample_bynode': 1, \n",
    "    'colsample_bytree': 1, \n",
    "    'learning_rate': 0.05, \n",
    "    'max_bin': 127, \n",
    "    'max_depth': 10, \n",
    "    'min_child_samples': 20, \n",
    "    'min_data_in_bin': 3, \n",
    "    'n_estimators': 300, \n",
    "    'num_leaves': 31, \n",
    "    'path_smooth': 0.2, \n",
    "    'subsample': 1\n",
    "}\n",
    "random_state = 0\n",
    "score_avg = 'macro'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "875035db",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "LGBMClassifier(class_weight='balanced', colsample_bynode=1, colsample_bytree=1,\n",
       "               learning_rate=0.05, max_bin=127, max_depth=10, min_data_in_bin=3,\n",
       "               n_estimators=300, path_smooth=0.2, random_state=0, subsample=1,\n",
       "               tree_learner='feature')"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from lightgbm import LGBMClassifier\n",
    "\n",
    "model = LGBMClassifier(\n",
    "    tree_learner='feature', \n",
    "    class_weight='balanced',\n",
    "    random_state=random_state,\n",
    ")\n",
    "model.set_params(**param_dict)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "6b556fef",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "LGBMClassifier(class_weight='balanced', colsample_bynode=1, colsample_bytree=1,\n",
       "               learning_rate=0.05, max_bin=127, max_depth=10, min_data_in_bin=3,\n",
       "               n_estimators=300, path_smooth=0.2, random_state=0, subsample=1,\n",
       "               tree_learner='feature')"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model.fit(X_train_norm, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "id": "a27f627f",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<AxesSubplot:xlabel='Feature'>"
      ]
     },
     "execution_count": 24,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x360 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "%matplotlib inline\n",
    "\n",
    "f_dict = {\n",
    "    'Feature': X_train.columns,\n",
    "    'Importance': model.feature_importances_\n",
    "}\n",
    "feature_df = pd.DataFrame(data=f_dict).sort_values(by='Importance', ascending=False)\n",
    "feature_df = feature_df[feature_df.Importance >= 1000]\n",
    "plt.rcParams[\"figure.figsize\"] = (10, 5)\n",
    "feature_df.set_index('Feature').plot.bar(rot=45)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "daad687f",
   "metadata": {},
   "source": [
    "## 5.2 Test Data Predictions"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "id": "fa0c63c8",
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "from sklearn.metrics import roc_auc_score, balanced_accuracy_score, f1_score, precision_score, recall_score, cohen_kappa_score\n",
    "from sklearn.metrics import classification_report\n",
    "\n",
    "y_pred = model.predict(X_test_norm)\n",
    "y_pred_proba = model.predict_proba(X_test)\n",
    "labels = np.unique(y_pred)\n",
    "test_auc = roc_auc_score(y_test, y_pred_proba, labels=labels, average=score_avg, multi_class='ovr')\n",
    "test_f1 = f1_score(y_test, y_pred, labels=labels, average=score_avg)\n",
    "test_precision = precision_score(y_test, y_pred, labels=labels, average=score_avg)\n",
    "test_recall = recall_score(y_test, y_pred, labels=labels, average=score_avg)\n",
    "test_ba = balanced_accuracy_score(y_test, y_pred)\n",
    "test_kappa = cohen_kappa_score(y_test, y_pred, labels=labels)\n",
    "results_dict = {\n",
    "    'ROC_AUC': [test_auc],\n",
    "    'F1 Score': [test_f1],\n",
    "    'Precision': [test_precision],\n",
    "    'Recall': [test_recall],\n",
    "    'Balanced Accuracy': [test_ba],\n",
    "    'Kappa Score': [test_kappa]\n",
    "}\n",
    "results_df = pd.DataFrame(data=results_dict)\n",
    "report = classification_report(y_test, y_pred, target_names=labels.astype(str), labels=labels)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "id": "8ee6fd07",
   "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>0</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>ROC_AUC</th>\n",
       "      <td>0.622634</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>F1 Score</th>\n",
       "      <td>0.514768</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Precision</th>\n",
       "      <td>0.471515</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Recall</th>\n",
       "      <td>0.633385</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Balanced Accuracy</th>\n",
       "      <td>0.633385</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Kappa Score</th>\n",
       "      <td>0.329062</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                          0\n",
       "ROC_AUC            0.622634\n",
       "F1 Score           0.514768\n",
       "Precision          0.471515\n",
       "Recall             0.633385\n",
       "Balanced Accuracy  0.633385\n",
       "Kappa Score        0.329062"
      ]
     },
     "execution_count": 30,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "results_df.T"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 411,
   "id": "939337f6",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "           1       0.50      0.66      0.57       584\n",
      "           2       0.36      0.67      0.47       354\n",
      "           3       0.55      0.81      0.65       566\n",
      "           4       0.50      1.00      0.67         2\n",
      "           5       0.64      0.30      0.41      3418\n",
      "           6       0.49      0.47      0.48      2506\n",
      "           7       0.39      0.54      0.45       527\n",
      "           8       0.15      0.48      0.23       164\n",
      "           9       0.15      0.41      0.22       370\n",
      "          10       0.97      1.00      0.99        76\n",
      "\n",
      "    accuracy                           0.45      8567\n",
      "   macro avg       0.47      0.63      0.51      8567\n",
      "weighted avg       0.53      0.45      0.46      8567\n",
      "\n"
     ]
    }
   ],
   "source": [
    "print(report)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "id": "88d502da",
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.metrics import confusion_matrix, ConfusionMatrixDisplay\n",
    "import matplotlib.pyplot as plt\n",
    "%matplotlib inline\n",
    "\n",
    "def plot_confusion_matrix(cm, classes,\n",
    "                          normalize=False,\n",
    "                          title='Confusion matrix',\n",
    "                          cmap=plt.cm.Blues):\n",
    "    \"\"\"\n",
    "    This function prints and plots the confusion matrix.\n",
    "    Normalization can be applied by setting `normalize=True`.\n",
    "    \"\"\"\n",
    "    import itertools\n",
    "    if normalize:\n",
    "        cm = cm.astype('float') / cm.sum(axis=1)[:, np.newaxis]\n",
    "        print(\"Normalized confusion matrix\")\n",
    "\n",
    "\n",
    "    plt.imshow(cm, interpolation='nearest', cmap=cmap)\n",
    "    plt.title(title)\n",
    "    plt.colorbar()\n",
    "    tick_marks = np.arange(len(classes))\n",
    "    plt.xticks(tick_marks, classes, rotation=45)\n",
    "    plt.yticks(tick_marks, classes)\n",
    "\n",
    "    fmt = '.2f' if normalize else 'd'\n",
    "    thresh = cm.max() / 2.\n",
    "    for i, j in itertools.product(range(cm.shape[0]), range(cm.shape[1])):\n",
    "        plt.text(j, i, format(cm[i, j], fmt),\n",
    "                 horizontalalignment=\"center\",\n",
    "                 color=\"white\" if cm[i, j] > thresh else \"black\")\n",
    "\n",
    "    plt.ylabel('True label')\n",
    "    plt.xlabel('Predicted label')\n",
    "    plt.tight_layout()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "id": "6c98924f",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1440x1440 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "cnf_matrix = confusion_matrix(y_test, y_pred,labels=model.classes_)\n",
    "np.set_printoptions(precision=2)\n",
    "\n",
    "# Plot non-normalized confusion matrix\n",
    "plt.rcParams[\"figure.figsize\"] = (20, 20)\n",
    "plt.rcParams[\"font.size\"] = 18\n",
    "plt.figure()\n",
    "plot_confusion_matrix(cnf_matrix, classes=model.classes_)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "949c26df",
   "metadata": {},
   "source": [
    "# 6. Summary"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "2a1d5e6b",
   "metadata": {},
   "source": [
    "We developed a machine learning-based predictive model to predict the vehicle emission score or smog rating (vehicle tailpipe CO2 emission) as defined by the Environmental Protection Agency (EPA). Essentially, the task is to perform a multi-class classification of the emission score, a discrete unitless integer quantity, varying from 1 (worst) to 10 (best). For this purpose, we collected the ‘vehicle’ and ‘emissions’ data sets from the EPA portal [1], [2]. \n",
    "\n",
    "As of now, the ‘vehicle’ and ‘emissions’ data contain around 44000 and 42000 records each with 83 and 8 features or columns, respectively. However, these data sets must be joined based on the vehicle id column which reduces the number of available observations because of missing ids. Moreover, the emission score had incorrect entries (negative scores) which were removed. Hence, after the data merging and cleaning steps, we had around 28000 observations. \n",
    "\n",
    "Thereafter, for categorical data transformation, we aggregated the fuelType1 column by combining midgrade and regular gasolines as 'Regular Gasoline' and 'Electricity' and 'Natural Gas' as 'Green Fuel'. Next, we performed correlation and distribution analysis, using which we were able to answer one of the sub research questions. We found that Standard Pickup Trucks and Vans (both cargo and passenger types) had the lowest emission scores in general. In addition, it was confirmed from boxplot analysis that, diesel vehicles received the lowest scores. \n",
    "\n",
    "An interesting observation from our boxplot outlier analysis was that all scores except those in the interval [4, 7] were marked as outliers. Hence, our initial choice of not performing outlier removal as part of the second milestone was justifiable. Similarly, we found principal component analysis (PCA) to be of not much use possibly because of strong non-linearity in the data set. We may need to apply kernel PCA-based approaches for appropriately investigating the benefits of PCA.\n",
    "\n",
    "The final component of this project was the predictive modeling where we compared seven classification models. These are Multinomial Logistic Regression (MLR), Support Vector Machine (SVM), Gradient Boosting Trees (GBT), Random Forests (RF), Boosted Random Forests (BRF), Extremely Randomized Trees (ETR), and Multi-Layer Perceptron (MLP). For the GBT and BRF algorithms, we used the LightGBM [3] package and used scikit-learn for the rest. \n",
    "\n",
    "One issue with the given data set is extremely high class imbalance. For example, the EPA score 4 only has 6 samples (0.02%) in total while 40% of the total emission scores are marked as 5. Another potential problem was the data dimensionality (35 predictors in total including dummies from the VClass column). \n",
    "\n",
    "Therefore, we preferred using ensemble machine learning algorithms that support class weights as parameters and provide feature selection on the fly. Moreover, we only have 28000 observations, so going for deep learning models is overkill. \n",
    "\n",
    "In order to appropriately compare all the models, we scaled the features in the [0, 1] interval using MinMaxScaler. Next, to train the models, we initially used the scikit-learn GridSearchCV with RepeatedKFold having five-folds and a single repeat (we tried multiple repeats too, but the results were similar). Here, we could not use the stratified version because class 4 only had four observations in the training data. This was also a reason why we used 70-30 split of the training and test data. If we used 80% as training, then class 4 only had a single observation in the test data. \n",
    "\n",
    "The model evaluation was based on the similarity of the training and validation F1 score to reduce overfitting. We then checked the F1 for the test data. For additional interpretability, we also showed balanced accuracy, precision, recall, and the Cohen’s kappa score. Note that F1 was the metric that was optimized. For, F1, precision, and recall, we used the macro averaging as part of the GridSearch scoring metrics. In this context, we had NaN issues with the ROC AUC macro OVR (one-versus-rest) scoring metric through GridSearchCV. That is why we only show the ROC AUC for the test data of the final model.\n",
    "\n",
    "While running the scikit-learn GridSearchCV with different hyperparameters of the ensemble learning algorithms and MLP on an Alienware M17 R1 (personal PC), we noticed that for many parameter values, even 8 hrs of computation was not sufficient. As a result, we switched to the Missouri S&T Foundry High-Performance Computing cluster and used Dask [4] to train our models in a distributed computing environment. This resulted in a significant reduction of computing time wherein we could tune several parameters in a very short time. For example, the GBM model training took around 6 hrs on the Alienware and took only 15 minutes using Dask.\n",
    "\n",
    "\n",
    "Finally, we found that the GBM shown in section 5 has the best performance in terms of all the error metrics for the test data."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "db943f91",
   "metadata": {},
   "source": [
    "# References"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f1c68610",
   "metadata": {},
   "source": [
    "[1]\tEPA, “EPA Fuel Economy,” 2021. https://www.fueleconomy.gov/ (accessed Dec. 05, 2021).\n",
    "\n",
    "[2]\tEPA, “Fuel Economy Guide,” 2021. [Online]. Available: https://www.fueleconomy.gov/feg/pdfs/guides/FEG2021.pdf.\n",
    "\n",
    "[3]\tG. Ke et al., “LightGBM: A Highly Efficient Gradient Boosting Decision Tree,” in Advances in Neural Information Processing Systems, 2017, vol. 30, [Online]. Available: https://proceedings.neurips.cc/paper/2017/file/6449f44a102fde848669bdd9eb6b76fa-Paper.pdf.\n",
    "\n",
    "[4]\tM. Rocklin, “Dask: Parallel Computation with Blocked algorithms and Task Scheduling,” Proc. 14th Python Sci. Conf, 2015, Accessed: 29-Nov-2021. [Online]. Available: http://conference.scipy.org/proceedings/scipy2015/pdfs/matthew_rocklin.pdf.\n"
   ]
  }
 ],
 "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": "3.9.4"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}