09. Predicting Car Prices.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Predicting car market price with KNN\n",
    "\n",
    "We are tasked with predicting what certain cars will be sold for on the market, depending on their attributes. The dataset is taken from the UCI Machine Learning Repository [here](https://archive.ics.uci.edu/ml/datasets/automobile).\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Cleaning"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "col_names = ['symboling','normalized_losses','make','fuel_type','aspiration',\n",
    "             'num_of_doors','body_style','drive_wheels','engine_location',\n",
    "             'wheel_base','length','width','height','curb_weight',\n",
    "             'engine_type','num_cyl','engine_size','fuel_system','bore',\n",
    "             'stroke','compression_ratio','horsepower','peak_rpm',\n",
    "             'city_mpg','highway_mpg','price']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "collapsed": false,
    "jupyter": {
     "outputs_hidden": false
    }
   },
   "outputs": [],
   "source": [
    "cars = pd.read_csv('imports-85.data', header=None)\n",
    "cars.columns = col_names"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "collapsed": false,
    "jupyter": {
     "outputs_hidden": false
    }
   },
   "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>symboling</th>\n",
       "      <th>normalized_losses</th>\n",
       "      <th>make</th>\n",
       "      <th>fuel_type</th>\n",
       "      <th>aspiration</th>\n",
       "      <th>num_of_doors</th>\n",
       "      <th>body_style</th>\n",
       "      <th>drive_wheels</th>\n",
       "      <th>engine_location</th>\n",
       "      <th>wheel_base</th>\n",
       "      <th>...</th>\n",
       "      <th>engine_size</th>\n",
       "      <th>fuel_system</th>\n",
       "      <th>bore</th>\n",
       "      <th>stroke</th>\n",
       "      <th>compression_ratio</th>\n",
       "      <th>horsepower</th>\n",
       "      <th>peak_rpm</th>\n",
       "      <th>city_mpg</th>\n",
       "      <th>highway_mpg</th>\n",
       "      <th>price</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>3</td>\n",
       "      <td>?</td>\n",
       "      <td>alfa-romero</td>\n",
       "      <td>gas</td>\n",
       "      <td>std</td>\n",
       "      <td>two</td>\n",
       "      <td>convertible</td>\n",
       "      <td>rwd</td>\n",
       "      <td>front</td>\n",
       "      <td>88.6</td>\n",
       "      <td>...</td>\n",
       "      <td>130</td>\n",
       "      <td>mpfi</td>\n",
       "      <td>3.47</td>\n",
       "      <td>2.68</td>\n",
       "      <td>9.0</td>\n",
       "      <td>111</td>\n",
       "      <td>5000</td>\n",
       "      <td>21</td>\n",
       "      <td>27</td>\n",
       "      <td>13495</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>3</td>\n",
       "      <td>?</td>\n",
       "      <td>alfa-romero</td>\n",
       "      <td>gas</td>\n",
       "      <td>std</td>\n",
       "      <td>two</td>\n",
       "      <td>convertible</td>\n",
       "      <td>rwd</td>\n",
       "      <td>front</td>\n",
       "      <td>88.6</td>\n",
       "      <td>...</td>\n",
       "      <td>130</td>\n",
       "      <td>mpfi</td>\n",
       "      <td>3.47</td>\n",
       "      <td>2.68</td>\n",
       "      <td>9.0</td>\n",
       "      <td>111</td>\n",
       "      <td>5000</td>\n",
       "      <td>21</td>\n",
       "      <td>27</td>\n",
       "      <td>16500</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>1</td>\n",
       "      <td>?</td>\n",
       "      <td>alfa-romero</td>\n",
       "      <td>gas</td>\n",
       "      <td>std</td>\n",
       "      <td>two</td>\n",
       "      <td>hatchback</td>\n",
       "      <td>rwd</td>\n",
       "      <td>front</td>\n",
       "      <td>94.5</td>\n",
       "      <td>...</td>\n",
       "      <td>152</td>\n",
       "      <td>mpfi</td>\n",
       "      <td>2.68</td>\n",
       "      <td>3.47</td>\n",
       "      <td>9.0</td>\n",
       "      <td>154</td>\n",
       "      <td>5000</td>\n",
       "      <td>19</td>\n",
       "      <td>26</td>\n",
       "      <td>16500</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>2</td>\n",
       "      <td>164</td>\n",
       "      <td>audi</td>\n",
       "      <td>gas</td>\n",
       "      <td>std</td>\n",
       "      <td>four</td>\n",
       "      <td>sedan</td>\n",
       "      <td>fwd</td>\n",
       "      <td>front</td>\n",
       "      <td>99.8</td>\n",
       "      <td>...</td>\n",
       "      <td>109</td>\n",
       "      <td>mpfi</td>\n",
       "      <td>3.19</td>\n",
       "      <td>3.40</td>\n",
       "      <td>10.0</td>\n",
       "      <td>102</td>\n",
       "      <td>5500</td>\n",
       "      <td>24</td>\n",
       "      <td>30</td>\n",
       "      <td>13950</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>2</td>\n",
       "      <td>164</td>\n",
       "      <td>audi</td>\n",
       "      <td>gas</td>\n",
       "      <td>std</td>\n",
       "      <td>four</td>\n",
       "      <td>sedan</td>\n",
       "      <td>4wd</td>\n",
       "      <td>front</td>\n",
       "      <td>99.4</td>\n",
       "      <td>...</td>\n",
       "      <td>136</td>\n",
       "      <td>mpfi</td>\n",
       "      <td>3.19</td>\n",
       "      <td>3.40</td>\n",
       "      <td>8.0</td>\n",
       "      <td>115</td>\n",
       "      <td>5500</td>\n",
       "      <td>18</td>\n",
       "      <td>22</td>\n",
       "      <td>17450</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 26 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "   symboling normalized_losses         make fuel_type aspiration num_of_doors  \\\n",
       "0          3                 ?  alfa-romero       gas        std          two   \n",
       "1          3                 ?  alfa-romero       gas        std          two   \n",
       "2          1                 ?  alfa-romero       gas        std          two   \n",
       "3          2               164         audi       gas        std         four   \n",
       "4          2               164         audi       gas        std         four   \n",
       "\n",
       "    body_style drive_wheels engine_location  wheel_base  ...  engine_size  \\\n",
       "0  convertible          rwd           front        88.6  ...          130   \n",
       "1  convertible          rwd           front        88.6  ...          130   \n",
       "2    hatchback          rwd           front        94.5  ...          152   \n",
       "3        sedan          fwd           front        99.8  ...          109   \n",
       "4        sedan          4wd           front        99.4  ...          136   \n",
       "\n",
       "   fuel_system  bore  stroke compression_ratio horsepower  peak_rpm city_mpg  \\\n",
       "0         mpfi  3.47    2.68               9.0        111      5000       21   \n",
       "1         mpfi  3.47    2.68               9.0        111      5000       21   \n",
       "2         mpfi  2.68    3.47               9.0        154      5000       19   \n",
       "3         mpfi  3.19    3.40              10.0        102      5500       24   \n",
       "4         mpfi  3.19    3.40               8.0        115      5500       18   \n",
       "\n",
       "  highway_mpg  price  \n",
       "0          27  13495  \n",
       "1          27  16500  \n",
       "2          26  16500  \n",
       "3          30  13950  \n",
       "4          22  17450  \n",
       "\n",
       "[5 rows x 26 columns]"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "cars.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "collapsed": false,
    "jupyter": {
     "outputs_hidden": false
    }
   },
   "outputs": [],
   "source": [
    "cars = cars.replace('?',np.nan)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "collapsed": false,
    "jupyter": {
     "outputs_hidden": false
    }
   },
   "outputs": [],
   "source": [
    "cars = cars.astype({\n",
    "    'normalized_losses':float,\n",
    "    'bore':float,\n",
    "    'stroke':float,\n",
    "    'horsepower':float,\n",
    "    'peak_rpm':float,\n",
    "    'price':float\n",
    "})"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "KNN algorithms can only use continuous variables with all observations having a value, so we start by filtering selecting only the appropriate columns. According to the data dictionary the following columns are continuous:\n",
    "\n",
    "* `normalized_losses`\n",
    "* `wheel_base`\n",
    "* `length`\n",
    "* `width`\n",
    "* `height`\n",
    "* `curb_weight`\n",
    "* `engine_size`\n",
    "* `bore`\n",
    "* `stroke`\n",
    "* `compression-ratio`\n",
    "* `horsepower`\n",
    "* `peak_rpm`\n",
    "* `city_mpg`\n",
    "* `highway_mpg`\n",
    "* `price`"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "num_cars = cars[['make','normalized_losses','wheel_base',\n",
    "                 'length','width','height','curb_weight',\n",
    "                 'engine_size','bore','stroke','compression_ratio','horsepower',\n",
    "                 'peak_rpm','city_mpg','highway_mpg','price']].copy()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Nan values"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "RangeIndex: 205 entries, 0 to 204\n",
      "Data columns (total 16 columns):\n",
      " #   Column             Non-Null Count  Dtype  \n",
      "---  ------             --------------  -----  \n",
      " 0   make               205 non-null    object \n",
      " 1   normalized_losses  164 non-null    float64\n",
      " 2   wheel_base         205 non-null    float64\n",
      " 3   length             205 non-null    float64\n",
      " 4   width              205 non-null    float64\n",
      " 5   height             205 non-null    float64\n",
      " 6   curb_weight        205 non-null    int64  \n",
      " 7   engine_size        205 non-null    int64  \n",
      " 8   bore               201 non-null    float64\n",
      " 9   stroke             201 non-null    float64\n",
      " 10  compression_ratio  205 non-null    float64\n",
      " 11  horsepower         203 non-null    float64\n",
      " 12  peak_rpm           203 non-null    float64\n",
      " 13  city_mpg           205 non-null    int64  \n",
      " 14  highway_mpg        205 non-null    int64  \n",
      " 15  price              201 non-null    float64\n",
      "dtypes: float64(11), int64(4), object(1)\n",
      "memory usage: 25.8+ KB\n"
     ]
    }
   ],
   "source": [
    "num_cars.info()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "make                  0\n",
       "normalized_losses    41\n",
       "wheel_base            0\n",
       "length                0\n",
       "width                 0\n",
       "height                0\n",
       "curb_weight           0\n",
       "engine_size           0\n",
       "bore                  4\n",
       "stroke                4\n",
       "compression_ratio     0\n",
       "horsepower            2\n",
       "peak_rpm              2\n",
       "city_mpg              0\n",
       "highway_mpg           0\n",
       "price                 4\n",
       "dtype: int64"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "num_cars.isna().sum()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The following columns contain missing values:\n",
    "\n",
    "* `normalized_losses`\n",
    "* `bore`\n",
    "* `stroke`\n",
    "* `horsepower`\n",
    "* `peak_rpm`\n",
    "* `price`"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### price"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "All four cars without a price are also missing the `normalized_losses` variable. We cannot have NaN as a `price` for our model. We have two choices:\n",
    "\n",
    "* Remove these four rows\n",
    "* Replace the NaN with the average of the prices"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "nan_price = num_cars[num_cars['price'].isna()]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "make\n",
       "audi       0\n",
       "isuzu      0\n",
       "porsche    0\n",
       "Name: price, dtype: int64"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "nan_price.groupby('make')['price'].count()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Since the price depends on as of yet unknown features, we will remove the four rows to avoid biasing the model when predicting prices for these three makes."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "num_cars = num_cars.dropna(subset=['price'],axis = 0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "collapsed": false,
    "jupyter": {
     "outputs_hidden": false
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "num_cars['price'].isna().sum()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### normalized_losses"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The `normalized_losses` feature has the most `nan` values and is worth investigating."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "collapsed": false,
    "jupyter": {
     "outputs_hidden": false
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "37"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "num_cars['normalized_losses'].isna().sum()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "collapsed": false,
    "jupyter": {
     "outputs_hidden": false
    }
   },
   "outputs": [],
   "source": [
    "# count the number of cars per maker\n",
    "car_count = pd.DataFrame(num_cars.groupby('make').count()['length']).reset_index()\n",
    "car_count.columns=['make','obs_num']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "collapsed": false,
    "jupyter": {
     "outputs_hidden": false
    }
   },
   "outputs": [],
   "source": [
    "# count the number of nan values per maker\n",
    "na_norm_losses_count = pd.DataFrame(num_cars[num_cars['normalized_losses'].isna()]['make'].value_counts()).reset_index()\n",
    "na_norm_losses_count.columns = ['make','nan_num']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {
    "collapsed": false,
    "jupyter": {
     "outputs_hidden": false
    }
   },
   "outputs": [],
   "source": [
    "# merge the two datasets and calculate the difference\n",
    "compare = car_count.merge(na_norm_losses_count)\n",
    "compare['no_nan'] = compare['obs_num'] - compare['nan_num']\n",
    "compare['perc_nan'] = round((compare['nan_num']/compare['obs_num'])*100,2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {
    "collapsed": false,
    "jupyter": {
     "outputs_hidden": false
    }
   },
   "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>make</th>\n",
       "      <th>obs_num</th>\n",
       "      <th>nan_num</th>\n",
       "      <th>no_nan</th>\n",
       "      <th>perc_nan</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>alfa-romero</td>\n",
       "      <td>3</td>\n",
       "      <td>3</td>\n",
       "      <td>0</td>\n",
       "      <td>100.00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>isuzu</td>\n",
       "      <td>2</td>\n",
       "      <td>2</td>\n",
       "      <td>0</td>\n",
       "      <td>100.00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>mercury</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>100.00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td>renault</td>\n",
       "      <td>2</td>\n",
       "      <td>2</td>\n",
       "      <td>0</td>\n",
       "      <td>100.00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>porsche</td>\n",
       "      <td>4</td>\n",
       "      <td>3</td>\n",
       "      <td>1</td>\n",
       "      <td>75.00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>jaguar</td>\n",
       "      <td>3</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>66.67</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>bmw</td>\n",
       "      <td>8</td>\n",
       "      <td>4</td>\n",
       "      <td>4</td>\n",
       "      <td>50.00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>mercedes-benz</td>\n",
       "      <td>8</td>\n",
       "      <td>3</td>\n",
       "      <td>5</td>\n",
       "      <td>37.50</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>peugot</td>\n",
       "      <td>11</td>\n",
       "      <td>4</td>\n",
       "      <td>7</td>\n",
       "      <td>36.36</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>audi</td>\n",
       "      <td>6</td>\n",
       "      <td>2</td>\n",
       "      <td>4</td>\n",
       "      <td>33.33</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14</th>\n",
       "      <td>volkswagen</td>\n",
       "      <td>12</td>\n",
       "      <td>4</td>\n",
       "      <td>8</td>\n",
       "      <td>33.33</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>mitsubishi</td>\n",
       "      <td>13</td>\n",
       "      <td>3</td>\n",
       "      <td>10</td>\n",
       "      <td>23.08</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>plymouth</td>\n",
       "      <td>7</td>\n",
       "      <td>1</td>\n",
       "      <td>6</td>\n",
       "      <td>14.29</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>mazda</td>\n",
       "      <td>17</td>\n",
       "      <td>2</td>\n",
       "      <td>15</td>\n",
       "      <td>11.76</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13</th>\n",
       "      <td>toyota</td>\n",
       "      <td>32</td>\n",
       "      <td>1</td>\n",
       "      <td>31</td>\n",
       "      <td>3.12</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "             make  obs_num  nan_num  no_nan  perc_nan\n",
       "0     alfa-romero        3        3       0    100.00\n",
       "3           isuzu        2        2       0    100.00\n",
       "7         mercury        1        1       0    100.00\n",
       "12        renault        2        2       0    100.00\n",
       "11        porsche        4        3       1     75.00\n",
       "4          jaguar        3        2       1     66.67\n",
       "2             bmw        8        4       4     50.00\n",
       "6   mercedes-benz        8        3       5     37.50\n",
       "9          peugot       11        4       7     36.36\n",
       "1            audi        6        2       4     33.33\n",
       "14     volkswagen       12        4       8     33.33\n",
       "8      mitsubishi       13        3      10     23.08\n",
       "10       plymouth        7        1       6     14.29\n",
       "5           mazda       17        2      15     11.76\n",
       "13         toyota       32        1      31      3.12"
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "compare.sort_values('perc_nan', ascending=False)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "37 of the `normalized_losses` values are `NaN`. To fix this we will do the following:\n",
    "\n",
    "* Car makes who's `normalized_losses` are always `NaN` are removed. These makes coincidentally do not have many data to begin with (such as price), so not much is lost.\n",
    "    * `alfa-romero`\n",
    "    * `isuzu`\n",
    "    * `mercury`\n",
    "    * `renault`\n",
    "* Car makes who have some numbers for `normalized_losses` will be kept, and the `NaN` replaced with the average.\n",
    "\n",
    "According to the data dictionary the `normalized_losses` variable represents \"average loss payment per insured vehicle year\". It depends on various size classifications of the car, such as number of doors, sedan vs hatchback, and so on. We assume the following describes the vehicle's size:\n",
    "\n",
    "* `num_of_doors`\n",
    "* `body_style`\n",
    "* `length`\n",
    "* `width`\n",
    "* `height`\n",
    "* `curb_weight`\n",
    "* `engine_size`\n",
    "\n",
    "Of these variables `num_of_doors` and`body_style` are not continuous variables and so are not considered."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "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>make</th>\n",
       "      <th>normalized_losses</th>\n",
       "      <th>length</th>\n",
       "      <th>width</th>\n",
       "      <th>height</th>\n",
       "      <th>curb_weight</th>\n",
       "      <th>engine_size</th>\n",
       "      <th>price</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>alfa-romero</td>\n",
       "      <td>NaN</td>\n",
       "      <td>168.8</td>\n",
       "      <td>64.1</td>\n",
       "      <td>48.8</td>\n",
       "      <td>2548</td>\n",
       "      <td>130</td>\n",
       "      <td>13495.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>alfa-romero</td>\n",
       "      <td>NaN</td>\n",
       "      <td>168.8</td>\n",
       "      <td>64.1</td>\n",
       "      <td>48.8</td>\n",
       "      <td>2548</td>\n",
       "      <td>130</td>\n",
       "      <td>16500.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>alfa-romero</td>\n",
       "      <td>NaN</td>\n",
       "      <td>171.2</td>\n",
       "      <td>65.5</td>\n",
       "      <td>52.4</td>\n",
       "      <td>2823</td>\n",
       "      <td>152</td>\n",
       "      <td>16500.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>audi</td>\n",
       "      <td>NaN</td>\n",
       "      <td>177.3</td>\n",
       "      <td>66.3</td>\n",
       "      <td>53.1</td>\n",
       "      <td>2507</td>\n",
       "      <td>136</td>\n",
       "      <td>15250.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>audi</td>\n",
       "      <td>NaN</td>\n",
       "      <td>192.7</td>\n",
       "      <td>71.4</td>\n",
       "      <td>55.7</td>\n",
       "      <td>2954</td>\n",
       "      <td>136</td>\n",
       "      <td>18920.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "          make  normalized_losses  length  width  height  curb_weight  \\\n",
       "0  alfa-romero                NaN   168.8   64.1    48.8         2548   \n",
       "1  alfa-romero                NaN   168.8   64.1    48.8         2548   \n",
       "2  alfa-romero                NaN   171.2   65.5    52.4         2823   \n",
       "5         audi                NaN   177.3   66.3    53.1         2507   \n",
       "7         audi                NaN   192.7   71.4    55.7         2954   \n",
       "\n",
       "   engine_size    price  \n",
       "0          130  13495.0  \n",
       "1          130  16500.0  \n",
       "2          152  16500.0  \n",
       "5          136  15250.0  \n",
       "7          136  18920.0  "
      ]
     },
     "execution_count": 19,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "norm_loss_cols = ['make','normalized_losses',\n",
    "                   'length', 'width', 'height', 'curb_weight',\n",
    "                   'engine_size', 'price']\n",
    "\n",
    "norm_loss_df = num_cars[norm_loss_cols][num_cars[norm_loss_cols]['normalized_losses'].isna()]\n",
    "norm_loss_df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "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>make</th>\n",
       "      <th>price</th>\n",
       "      <th>normalized_losses</th>\n",
       "      <th>variable</th>\n",
       "      <th>value</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>alfa-romero</td>\n",
       "      <td>13495.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>length</td>\n",
       "      <td>168.8</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>alfa-romero</td>\n",
       "      <td>16500.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>length</td>\n",
       "      <td>168.8</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>alfa-romero</td>\n",
       "      <td>16500.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>length</td>\n",
       "      <td>171.2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>audi</td>\n",
       "      <td>15250.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>length</td>\n",
       "      <td>177.3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>audi</td>\n",
       "      <td>18920.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>length</td>\n",
       "      <td>192.7</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "          make    price  normalized_losses variable  value\n",
       "0  alfa-romero  13495.0                NaN   length  168.8\n",
       "1  alfa-romero  16500.0                NaN   length  168.8\n",
       "2  alfa-romero  16500.0                NaN   length  171.2\n",
       "3         audi  15250.0                NaN   length  177.3\n",
       "4         audi  18920.0                NaN   length  192.7"
      ]
     },
     "execution_count": 20,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "norm_loss_df = norm_loss_df.melt(id_vars = ['make','price','normalized_losses'])\n",
    "norm_loss_df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<AxesSubplot:xlabel='variable', ylabel='value'>"
      ]
     },
     "execution_count": 21,
     "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='variable',y = 'value', data= norm_loss_df[norm_loss_df['variable'] != 'curb_weight'])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "There are not too many outliers in these variables, so we will take the mean to fill the `nan`."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "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>normalized_losses</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>make</th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>alfa-romero</th>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>audi</th>\n",
       "      <td>161.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>bmw</th>\n",
       "      <td>190.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>chevrolet</th>\n",
       "      <td>100.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>dodge</th>\n",
       "      <td>133.444444</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "             normalized_losses\n",
       "make                          \n",
       "alfa-romero                NaN\n",
       "audi                161.000000\n",
       "bmw                 190.000000\n",
       "chevrolet           100.000000\n",
       "dodge               133.444444"
      ]
     },
     "execution_count": 22,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# figure out the mean of each make's normalized_losses\n",
    "norm_loss_means = num_cars.groupby('make').mean()[['normalized_losses']]\n",
    "norm_loss_means.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {
    "collapsed": false,
    "jupyter": {
     "outputs_hidden": false
    }
   },
   "outputs": [],
   "source": [
    "# identify each nan, replace it with the mean of that maker's normalized_losses\n",
    "for i, row in num_cars.iterrows():\n",
    "    if np.isnan(row['normalized_losses']):\n",
    "        make_mean = norm_loss_means.loc[row['make']][0]\n",
    "        num_cars.loc[i,'normalized_losses'] = make_mean"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "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>make</th>\n",
       "      <th>normalized_losses</th>\n",
       "      <th>wheel_base</th>\n",
       "      <th>length</th>\n",
       "      <th>width</th>\n",
       "      <th>height</th>\n",
       "      <th>curb_weight</th>\n",
       "      <th>engine_size</th>\n",
       "      <th>bore</th>\n",
       "      <th>stroke</th>\n",
       "      <th>compression_ratio</th>\n",
       "      <th>horsepower</th>\n",
       "      <th>peak_rpm</th>\n",
       "      <th>city_mpg</th>\n",
       "      <th>highway_mpg</th>\n",
       "      <th>price</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>alfa-romero</td>\n",
       "      <td>NaN</td>\n",
       "      <td>88.6</td>\n",
       "      <td>168.8</td>\n",
       "      <td>64.1</td>\n",
       "      <td>48.8</td>\n",
       "      <td>2548</td>\n",
       "      <td>130</td>\n",
       "      <td>3.47</td>\n",
       "      <td>2.68</td>\n",
       "      <td>9.0</td>\n",
       "      <td>111.0</td>\n",
       "      <td>5000.0</td>\n",
       "      <td>21</td>\n",
       "      <td>27</td>\n",
       "      <td>13495.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>alfa-romero</td>\n",
       "      <td>NaN</td>\n",
       "      <td>88.6</td>\n",
       "      <td>168.8</td>\n",
       "      <td>64.1</td>\n",
       "      <td>48.8</td>\n",
       "      <td>2548</td>\n",
       "      <td>130</td>\n",
       "      <td>3.47</td>\n",
       "      <td>2.68</td>\n",
       "      <td>9.0</td>\n",
       "      <td>111.0</td>\n",
       "      <td>5000.0</td>\n",
       "      <td>21</td>\n",
       "      <td>27</td>\n",
       "      <td>16500.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>alfa-romero</td>\n",
       "      <td>NaN</td>\n",
       "      <td>94.5</td>\n",
       "      <td>171.2</td>\n",
       "      <td>65.5</td>\n",
       "      <td>52.4</td>\n",
       "      <td>2823</td>\n",
       "      <td>152</td>\n",
       "      <td>2.68</td>\n",
       "      <td>3.47</td>\n",
       "      <td>9.0</td>\n",
       "      <td>154.0</td>\n",
       "      <td>5000.0</td>\n",
       "      <td>19</td>\n",
       "      <td>26</td>\n",
       "      <td>16500.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>43</th>\n",
       "      <td>isuzu</td>\n",
       "      <td>NaN</td>\n",
       "      <td>94.3</td>\n",
       "      <td>170.7</td>\n",
       "      <td>61.8</td>\n",
       "      <td>53.5</td>\n",
       "      <td>2337</td>\n",
       "      <td>111</td>\n",
       "      <td>3.31</td>\n",
       "      <td>3.23</td>\n",
       "      <td>8.5</td>\n",
       "      <td>78.0</td>\n",
       "      <td>4800.0</td>\n",
       "      <td>24</td>\n",
       "      <td>29</td>\n",
       "      <td>6785.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>46</th>\n",
       "      <td>isuzu</td>\n",
       "      <td>NaN</td>\n",
       "      <td>96.0</td>\n",
       "      <td>172.6</td>\n",
       "      <td>65.2</td>\n",
       "      <td>51.4</td>\n",
       "      <td>2734</td>\n",
       "      <td>119</td>\n",
       "      <td>3.43</td>\n",
       "      <td>3.23</td>\n",
       "      <td>9.2</td>\n",
       "      <td>90.0</td>\n",
       "      <td>5000.0</td>\n",
       "      <td>24</td>\n",
       "      <td>29</td>\n",
       "      <td>11048.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>75</th>\n",
       "      <td>mercury</td>\n",
       "      <td>NaN</td>\n",
       "      <td>102.7</td>\n",
       "      <td>178.4</td>\n",
       "      <td>68.0</td>\n",
       "      <td>54.8</td>\n",
       "      <td>2910</td>\n",
       "      <td>140</td>\n",
       "      <td>3.78</td>\n",
       "      <td>3.12</td>\n",
       "      <td>8.0</td>\n",
       "      <td>175.0</td>\n",
       "      <td>5000.0</td>\n",
       "      <td>19</td>\n",
       "      <td>24</td>\n",
       "      <td>16503.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>130</th>\n",
       "      <td>renault</td>\n",
       "      <td>NaN</td>\n",
       "      <td>96.1</td>\n",
       "      <td>181.5</td>\n",
       "      <td>66.5</td>\n",
       "      <td>55.2</td>\n",
       "      <td>2579</td>\n",
       "      <td>132</td>\n",
       "      <td>3.46</td>\n",
       "      <td>3.90</td>\n",
       "      <td>8.7</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>23</td>\n",
       "      <td>31</td>\n",
       "      <td>9295.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>131</th>\n",
       "      <td>renault</td>\n",
       "      <td>NaN</td>\n",
       "      <td>96.1</td>\n",
       "      <td>176.8</td>\n",
       "      <td>66.6</td>\n",
       "      <td>50.5</td>\n",
       "      <td>2460</td>\n",
       "      <td>132</td>\n",
       "      <td>3.46</td>\n",
       "      <td>3.90</td>\n",
       "      <td>8.7</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>23</td>\n",
       "      <td>31</td>\n",
       "      <td>9895.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "            make  normalized_losses  wheel_base  length  width  height  \\\n",
       "0    alfa-romero                NaN        88.6   168.8   64.1    48.8   \n",
       "1    alfa-romero                NaN        88.6   168.8   64.1    48.8   \n",
       "2    alfa-romero                NaN        94.5   171.2   65.5    52.4   \n",
       "43         isuzu                NaN        94.3   170.7   61.8    53.5   \n",
       "46         isuzu                NaN        96.0   172.6   65.2    51.4   \n",
       "75       mercury                NaN       102.7   178.4   68.0    54.8   \n",
       "130      renault                NaN        96.1   181.5   66.5    55.2   \n",
       "131      renault                NaN        96.1   176.8   66.6    50.5   \n",
       "\n",
       "     curb_weight  engine_size  bore  stroke  compression_ratio  horsepower  \\\n",
       "0           2548          130  3.47    2.68                9.0       111.0   \n",
       "1           2548          130  3.47    2.68                9.0       111.0   \n",
       "2           2823          152  2.68    3.47                9.0       154.0   \n",
       "43          2337          111  3.31    3.23                8.5        78.0   \n",
       "46          2734          119  3.43    3.23                9.2        90.0   \n",
       "75          2910          140  3.78    3.12                8.0       175.0   \n",
       "130         2579          132  3.46    3.90                8.7         NaN   \n",
       "131         2460          132  3.46    3.90                8.7         NaN   \n",
       "\n",
       "     peak_rpm  city_mpg  highway_mpg    price  \n",
       "0      5000.0        21           27  13495.0  \n",
       "1      5000.0        21           27  16500.0  \n",
       "2      5000.0        19           26  16500.0  \n",
       "43     4800.0        24           29   6785.0  \n",
       "46     5000.0        24           29  11048.0  \n",
       "75     5000.0        19           24  16503.0  \n",
       "130       NaN        23           31   9295.0  \n",
       "131       NaN        23           31   9895.0  "
      ]
     },
     "execution_count": 24,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "num_cars[num_cars['normalized_losses'].isna()]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We have 8 `nan` left. These makes do not have any `normalized_losses` calculated for any of there cars. Since there are only 8 of them, we will assign the overall mean."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [],
   "source": [
    "norm_loss_mean = num_cars['normalized_losses'].mean()\n",
    "num_cars['normalized_losses'] = num_cars['normalized_losses'].fillna(value = norm_loss_mean)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "make                 0\n",
       "normalized_losses    0\n",
       "wheel_base           0\n",
       "length               0\n",
       "width                0\n",
       "height               0\n",
       "curb_weight          0\n",
       "engine_size          0\n",
       "bore                 4\n",
       "stroke               4\n",
       "compression_ratio    0\n",
       "horsepower           2\n",
       "peak_rpm             2\n",
       "city_mpg             0\n",
       "highway_mpg          0\n",
       "price                0\n",
       "dtype: int64"
      ]
     },
     "execution_count": 26,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "num_cars.isna().sum()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "No more `nan` in the `normalized_losses` column!"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### bore and stroke"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {
    "collapsed": false,
    "jupyter": {
     "outputs_hidden": false
    }
   },
   "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>make</th>\n",
       "      <th>normalized_losses</th>\n",
       "      <th>wheel_base</th>\n",
       "      <th>length</th>\n",
       "      <th>width</th>\n",
       "      <th>height</th>\n",
       "      <th>curb_weight</th>\n",
       "      <th>engine_size</th>\n",
       "      <th>bore</th>\n",
       "      <th>stroke</th>\n",
       "      <th>compression_ratio</th>\n",
       "      <th>horsepower</th>\n",
       "      <th>peak_rpm</th>\n",
       "      <th>city_mpg</th>\n",
       "      <th>highway_mpg</th>\n",
       "      <th>price</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>55</th>\n",
       "      <td>mazda</td>\n",
       "      <td>150.0</td>\n",
       "      <td>95.3</td>\n",
       "      <td>169.0</td>\n",
       "      <td>65.7</td>\n",
       "      <td>49.6</td>\n",
       "      <td>2380</td>\n",
       "      <td>70</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>9.4</td>\n",
       "      <td>101.0</td>\n",
       "      <td>6000.0</td>\n",
       "      <td>17</td>\n",
       "      <td>23</td>\n",
       "      <td>10945.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>56</th>\n",
       "      <td>mazda</td>\n",
       "      <td>150.0</td>\n",
       "      <td>95.3</td>\n",
       "      <td>169.0</td>\n",
       "      <td>65.7</td>\n",
       "      <td>49.6</td>\n",
       "      <td>2380</td>\n",
       "      <td>70</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>9.4</td>\n",
       "      <td>101.0</td>\n",
       "      <td>6000.0</td>\n",
       "      <td>17</td>\n",
       "      <td>23</td>\n",
       "      <td>11845.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>57</th>\n",
       "      <td>mazda</td>\n",
       "      <td>150.0</td>\n",
       "      <td>95.3</td>\n",
       "      <td>169.0</td>\n",
       "      <td>65.7</td>\n",
       "      <td>49.6</td>\n",
       "      <td>2385</td>\n",
       "      <td>70</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>9.4</td>\n",
       "      <td>101.0</td>\n",
       "      <td>6000.0</td>\n",
       "      <td>17</td>\n",
       "      <td>23</td>\n",
       "      <td>13645.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>58</th>\n",
       "      <td>mazda</td>\n",
       "      <td>150.0</td>\n",
       "      <td>95.3</td>\n",
       "      <td>169.0</td>\n",
       "      <td>65.7</td>\n",
       "      <td>49.6</td>\n",
       "      <td>2500</td>\n",
       "      <td>80</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>9.4</td>\n",
       "      <td>135.0</td>\n",
       "      <td>6000.0</td>\n",
       "      <td>16</td>\n",
       "      <td>23</td>\n",
       "      <td>15645.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "     make  normalized_losses  wheel_base  length  width  height  curb_weight  \\\n",
       "55  mazda              150.0        95.3   169.0   65.7    49.6         2380   \n",
       "56  mazda              150.0        95.3   169.0   65.7    49.6         2380   \n",
       "57  mazda              150.0        95.3   169.0   65.7    49.6         2385   \n",
       "58  mazda              150.0        95.3   169.0   65.7    49.6         2500   \n",
       "\n",
       "    engine_size  bore  stroke  compression_ratio  horsepower  peak_rpm  \\\n",
       "55           70   NaN     NaN                9.4       101.0    6000.0   \n",
       "56           70   NaN     NaN                9.4       101.0    6000.0   \n",
       "57           70   NaN     NaN                9.4       101.0    6000.0   \n",
       "58           80   NaN     NaN                9.4       135.0    6000.0   \n",
       "\n",
       "    city_mpg  highway_mpg    price  \n",
       "55        17           23  10945.0  \n",
       "56        17           23  11845.0  \n",
       "57        17           23  13645.0  \n",
       "58        16           23  15645.0  "
      ]
     },
     "execution_count": 27,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "num_cars[num_cars['bore'].isna()]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "All the `nan` values for `bore` and `stroke` are under the `mazda` make, with a `hatchback` body style. \n",
    "\n",
    "We should avoid removing any more observations as the dataset is limited. We can note that `bore` and `stroke` are based on the type of the engine, and categorize the power it can make. As we have other variables to describe the engine (`engine_type`, `engine_size`) and the bore/stroke descriptions are summarized by the `compression_ratio`, we will remove these features."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {
    "collapsed": false,
    "jupyter": {
     "outputs_hidden": false
    }
   },
   "outputs": [],
   "source": [
    "num_cars = num_cars.drop(['bore','stroke'],axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "Int64Index: 201 entries, 0 to 204\n",
      "Data columns (total 14 columns):\n",
      " #   Column             Non-Null Count  Dtype  \n",
      "---  ------             --------------  -----  \n",
      " 0   make               201 non-null    object \n",
      " 1   normalized_losses  201 non-null    float64\n",
      " 2   wheel_base         201 non-null    float64\n",
      " 3   length             201 non-null    float64\n",
      " 4   width              201 non-null    float64\n",
      " 5   height             201 non-null    float64\n",
      " 6   curb_weight        201 non-null    int64  \n",
      " 7   engine_size        201 non-null    int64  \n",
      " 8   compression_ratio  201 non-null    float64\n",
      " 9   horsepower         199 non-null    float64\n",
      " 10  peak_rpm           199 non-null    float64\n",
      " 11  city_mpg           201 non-null    int64  \n",
      " 12  highway_mpg        201 non-null    int64  \n",
      " 13  price              201 non-null    float64\n",
      "dtypes: float64(9), int64(4), object(1)\n",
      "memory usage: 33.6+ KB\n"
     ]
    }
   ],
   "source": [
    "num_cars.info()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### horsepower and peak_rpm"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "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>make</th>\n",
       "      <th>normalized_losses</th>\n",
       "      <th>wheel_base</th>\n",
       "      <th>length</th>\n",
       "      <th>width</th>\n",
       "      <th>height</th>\n",
       "      <th>curb_weight</th>\n",
       "      <th>engine_size</th>\n",
       "      <th>compression_ratio</th>\n",
       "      <th>horsepower</th>\n",
       "      <th>peak_rpm</th>\n",
       "      <th>city_mpg</th>\n",
       "      <th>highway_mpg</th>\n",
       "      <th>price</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>130</th>\n",
       "      <td>renault</td>\n",
       "      <td>125.912731</td>\n",
       "      <td>96.1</td>\n",
       "      <td>181.5</td>\n",
       "      <td>66.5</td>\n",
       "      <td>55.2</td>\n",
       "      <td>2579</td>\n",
       "      <td>132</td>\n",
       "      <td>8.7</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>23</td>\n",
       "      <td>31</td>\n",
       "      <td>9295.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>131</th>\n",
       "      <td>renault</td>\n",
       "      <td>125.912731</td>\n",
       "      <td>96.1</td>\n",
       "      <td>176.8</td>\n",
       "      <td>66.6</td>\n",
       "      <td>50.5</td>\n",
       "      <td>2460</td>\n",
       "      <td>132</td>\n",
       "      <td>8.7</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>23</td>\n",
       "      <td>31</td>\n",
       "      <td>9895.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "        make  normalized_losses  wheel_base  length  width  height  \\\n",
       "130  renault         125.912731        96.1   181.5   66.5    55.2   \n",
       "131  renault         125.912731        96.1   176.8   66.6    50.5   \n",
       "\n",
       "     curb_weight  engine_size  compression_ratio  horsepower  peak_rpm  \\\n",
       "130         2579          132                8.7         NaN       NaN   \n",
       "131         2460          132                8.7         NaN       NaN   \n",
       "\n",
       "     city_mpg  highway_mpg   price  \n",
       "130        23           31  9295.0  \n",
       "131        23           31  9895.0  "
      ]
     },
     "execution_count": 30,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "num_cars[num_cars['horsepower'].isna()]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The two `nan` in these variables belong to the same make, and to observations with very similar characteristics"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "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>make</th>\n",
       "      <th>normalized_losses</th>\n",
       "      <th>wheel_base</th>\n",
       "      <th>length</th>\n",
       "      <th>width</th>\n",
       "      <th>height</th>\n",
       "      <th>curb_weight</th>\n",
       "      <th>engine_size</th>\n",
       "      <th>compression_ratio</th>\n",
       "      <th>horsepower</th>\n",
       "      <th>peak_rpm</th>\n",
       "      <th>city_mpg</th>\n",
       "      <th>highway_mpg</th>\n",
       "      <th>price</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>130</th>\n",
       "      <td>renault</td>\n",
       "      <td>125.912731</td>\n",
       "      <td>96.1</td>\n",
       "      <td>181.5</td>\n",
       "      <td>66.5</td>\n",
       "      <td>55.2</td>\n",
       "      <td>2579</td>\n",
       "      <td>132</td>\n",
       "      <td>8.7</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>23</td>\n",
       "      <td>31</td>\n",
       "      <td>9295.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>131</th>\n",
       "      <td>renault</td>\n",
       "      <td>125.912731</td>\n",
       "      <td>96.1</td>\n",
       "      <td>176.8</td>\n",
       "      <td>66.6</td>\n",
       "      <td>50.5</td>\n",
       "      <td>2460</td>\n",
       "      <td>132</td>\n",
       "      <td>8.7</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>23</td>\n",
       "      <td>31</td>\n",
       "      <td>9895.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "        make  normalized_losses  wheel_base  length  width  height  \\\n",
       "130  renault         125.912731        96.1   181.5   66.5    55.2   \n",
       "131  renault         125.912731        96.1   176.8   66.6    50.5   \n",
       "\n",
       "     curb_weight  engine_size  compression_ratio  horsepower  peak_rpm  \\\n",
       "130         2579          132                8.7         NaN       NaN   \n",
       "131         2460          132                8.7         NaN       NaN   \n",
       "\n",
       "     city_mpg  highway_mpg   price  \n",
       "130        23           31  9295.0  \n",
       "131        23           31  9895.0  "
      ]
     },
     "execution_count": 31,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "num_cars[num_cars['make']=='renault']"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "These two cars also happen to be the only `renault` observations, and the `normalized_losses` was estimated. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "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>make</th>\n",
       "      <th>normalized_losses</th>\n",
       "      <th>wheel_base</th>\n",
       "      <th>length</th>\n",
       "      <th>width</th>\n",
       "      <th>height</th>\n",
       "      <th>curb_weight</th>\n",
       "      <th>engine_size</th>\n",
       "      <th>compression_ratio</th>\n",
       "      <th>horsepower</th>\n",
       "      <th>peak_rpm</th>\n",
       "      <th>city_mpg</th>\n",
       "      <th>highway_mpg</th>\n",
       "      <th>price</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>100</th>\n",
       "      <td>nissan</td>\n",
       "      <td>106.000000</td>\n",
       "      <td>97.2</td>\n",
       "      <td>173.4</td>\n",
       "      <td>65.2</td>\n",
       "      <td>54.7</td>\n",
       "      <td>2302</td>\n",
       "      <td>120</td>\n",
       "      <td>8.5</td>\n",
       "      <td>97.0</td>\n",
       "      <td>5200.0</td>\n",
       "      <td>27</td>\n",
       "      <td>34</td>\n",
       "      <td>9549.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>130</th>\n",
       "      <td>renault</td>\n",
       "      <td>125.912731</td>\n",
       "      <td>96.1</td>\n",
       "      <td>181.5</td>\n",
       "      <td>66.5</td>\n",
       "      <td>55.2</td>\n",
       "      <td>2579</td>\n",
       "      <td>132</td>\n",
       "      <td>8.7</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>23</td>\n",
       "      <td>31</td>\n",
       "      <td>9295.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>131</th>\n",
       "      <td>renault</td>\n",
       "      <td>125.912731</td>\n",
       "      <td>96.1</td>\n",
       "      <td>176.8</td>\n",
       "      <td>66.6</td>\n",
       "      <td>50.5</td>\n",
       "      <td>2460</td>\n",
       "      <td>132</td>\n",
       "      <td>8.7</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>23</td>\n",
       "      <td>31</td>\n",
       "      <td>9895.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>175</th>\n",
       "      <td>toyota</td>\n",
       "      <td>65.000000</td>\n",
       "      <td>102.4</td>\n",
       "      <td>175.6</td>\n",
       "      <td>66.5</td>\n",
       "      <td>53.9</td>\n",
       "      <td>2414</td>\n",
       "      <td>122</td>\n",
       "      <td>8.7</td>\n",
       "      <td>92.0</td>\n",
       "      <td>4200.0</td>\n",
       "      <td>27</td>\n",
       "      <td>32</td>\n",
       "      <td>9988.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>190</th>\n",
       "      <td>volkswagen</td>\n",
       "      <td>256.000000</td>\n",
       "      <td>94.5</td>\n",
       "      <td>165.7</td>\n",
       "      <td>64.0</td>\n",
       "      <td>51.4</td>\n",
       "      <td>2221</td>\n",
       "      <td>109</td>\n",
       "      <td>8.5</td>\n",
       "      <td>90.0</td>\n",
       "      <td>5500.0</td>\n",
       "      <td>24</td>\n",
       "      <td>29</td>\n",
       "      <td>9980.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "           make  normalized_losses  wheel_base  length  width  height  \\\n",
       "100      nissan         106.000000        97.2   173.4   65.2    54.7   \n",
       "130     renault         125.912731        96.1   181.5   66.5    55.2   \n",
       "131     renault         125.912731        96.1   176.8   66.6    50.5   \n",
       "175      toyota          65.000000       102.4   175.6   66.5    53.9   \n",
       "190  volkswagen         256.000000        94.5   165.7   64.0    51.4   \n",
       "\n",
       "     curb_weight  engine_size  compression_ratio  horsepower  peak_rpm  \\\n",
       "100         2302          120                8.5        97.0    5200.0   \n",
       "130         2579          132                8.7         NaN       NaN   \n",
       "131         2460          132                8.7         NaN       NaN   \n",
       "175         2414          122                8.7        92.0    4200.0   \n",
       "190         2221          109                8.5        90.0    5500.0   \n",
       "\n",
       "     city_mpg  highway_mpg   price  \n",
       "100        27           34  9549.0  \n",
       "130        23           31  9295.0  \n",
       "131        23           31  9895.0  \n",
       "175        27           32  9988.0  \n",
       "190        24           29  9980.0  "
      ]
     },
     "execution_count": 32,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "renault_similars = num_cars[(num_cars['price']<10000) & (num_cars['price'] > 9000) & \n",
    "                            (num_cars['compression_ratio'] > 8) & (num_cars['compression_ratio'] < 9)]\n",
    "renault_similars"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The two renaults have very similar characteristics (of continuous variables at least) to 3 other cars. We will take the mean `horsepower` and `peak_rpm` of these 3 cars and use it for the renault."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [],
   "source": [
    "mean_horse = renault_similars[['horsepower','peak_rpm']].mean()[0]\n",
    "mean_rpm = renault_similars[['horsepower','peak_rpm']].mean()[1]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [],
   "source": [
    "for i in [130,131]:\n",
    "    num_cars.loc[i,'horsepower'] = mean_horse\n",
    "    num_cars.loc[i,'peak_rpm'] = mean_rpm"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "make                 0\n",
       "normalized_losses    0\n",
       "wheel_base           0\n",
       "length               0\n",
       "width                0\n",
       "height               0\n",
       "curb_weight          0\n",
       "engine_size          0\n",
       "compression_ratio    0\n",
       "horsepower           0\n",
       "peak_rpm             0\n",
       "city_mpg             0\n",
       "highway_mpg          0\n",
       "price                0\n",
       "dtype: int64"
      ]
     },
     "execution_count": 35,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "num_cars.isna().sum()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We have no more `nan` to take care of!"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Normalization"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "In order to properly compare across numeric features, we will normalize the values using min-max normalization:\n",
    "\n",
    "`(x - min(x))/(max(x) - min(x))`"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [],
   "source": [
    "norm_cars = num_cars.copy()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [],
   "source": [
    "norm_cols = list(norm_cars.columns)\n",
    "norm_cols.remove('price')\n",
    "norm_cols.remove('make')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {},
   "outputs": [],
   "source": [
    "# normalize everything except the price (and the make)\n",
    "for col in norm_cols:\n",
    "    norm_cars[col] = (norm_cars[col]-norm_cars[col].min())/(norm_cars[col].max()-norm_cars[col].min())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {
    "collapsed": false,
    "jupyter": {
     "outputs_hidden": false
    }
   },
   "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>make</th>\n",
       "      <th>normalized_losses</th>\n",
       "      <th>wheel_base</th>\n",
       "      <th>length</th>\n",
       "      <th>width</th>\n",
       "      <th>height</th>\n",
       "      <th>curb_weight</th>\n",
       "      <th>engine_size</th>\n",
       "      <th>compression_ratio</th>\n",
       "      <th>horsepower</th>\n",
       "      <th>peak_rpm</th>\n",
       "      <th>city_mpg</th>\n",
       "      <th>highway_mpg</th>\n",
       "      <th>price</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>alfa-romero</td>\n",
       "      <td>0.318915</td>\n",
       "      <td>0.058309</td>\n",
       "      <td>0.413433</td>\n",
       "      <td>0.324786</td>\n",
       "      <td>0.083333</td>\n",
       "      <td>0.411171</td>\n",
       "      <td>0.260377</td>\n",
       "      <td>0.1250</td>\n",
       "      <td>0.294393</td>\n",
       "      <td>0.346939</td>\n",
       "      <td>0.222222</td>\n",
       "      <td>0.289474</td>\n",
       "      <td>13495.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>alfa-romero</td>\n",
       "      <td>0.318915</td>\n",
       "      <td>0.058309</td>\n",
       "      <td>0.413433</td>\n",
       "      <td>0.324786</td>\n",
       "      <td>0.083333</td>\n",
       "      <td>0.411171</td>\n",
       "      <td>0.260377</td>\n",
       "      <td>0.1250</td>\n",
       "      <td>0.294393</td>\n",
       "      <td>0.346939</td>\n",
       "      <td>0.222222</td>\n",
       "      <td>0.289474</td>\n",
       "      <td>16500.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>alfa-romero</td>\n",
       "      <td>0.318915</td>\n",
       "      <td>0.230321</td>\n",
       "      <td>0.449254</td>\n",
       "      <td>0.444444</td>\n",
       "      <td>0.383333</td>\n",
       "      <td>0.517843</td>\n",
       "      <td>0.343396</td>\n",
       "      <td>0.1250</td>\n",
       "      <td>0.495327</td>\n",
       "      <td>0.346939</td>\n",
       "      <td>0.166667</td>\n",
       "      <td>0.263158</td>\n",
       "      <td>16500.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>audi</td>\n",
       "      <td>0.518325</td>\n",
       "      <td>0.384840</td>\n",
       "      <td>0.529851</td>\n",
       "      <td>0.504274</td>\n",
       "      <td>0.541667</td>\n",
       "      <td>0.329325</td>\n",
       "      <td>0.181132</td>\n",
       "      <td>0.1875</td>\n",
       "      <td>0.252336</td>\n",
       "      <td>0.551020</td>\n",
       "      <td>0.305556</td>\n",
       "      <td>0.368421</td>\n",
       "      <td>13950.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>audi</td>\n",
       "      <td>0.518325</td>\n",
       "      <td>0.373178</td>\n",
       "      <td>0.529851</td>\n",
       "      <td>0.521368</td>\n",
       "      <td>0.541667</td>\n",
       "      <td>0.518231</td>\n",
       "      <td>0.283019</td>\n",
       "      <td>0.0625</td>\n",
       "      <td>0.313084</td>\n",
       "      <td>0.551020</td>\n",
       "      <td>0.138889</td>\n",
       "      <td>0.157895</td>\n",
       "      <td>17450.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "          make  normalized_losses  wheel_base    length     width    height  \\\n",
       "0  alfa-romero           0.318915    0.058309  0.413433  0.324786  0.083333   \n",
       "1  alfa-romero           0.318915    0.058309  0.413433  0.324786  0.083333   \n",
       "2  alfa-romero           0.318915    0.230321  0.449254  0.444444  0.383333   \n",
       "3         audi           0.518325    0.384840  0.529851  0.504274  0.541667   \n",
       "4         audi           0.518325    0.373178  0.529851  0.521368  0.541667   \n",
       "\n",
       "   curb_weight  engine_size  compression_ratio  horsepower  peak_rpm  \\\n",
       "0     0.411171     0.260377             0.1250    0.294393  0.346939   \n",
       "1     0.411171     0.260377             0.1250    0.294393  0.346939   \n",
       "2     0.517843     0.343396             0.1250    0.495327  0.346939   \n",
       "3     0.329325     0.181132             0.1875    0.252336  0.551020   \n",
       "4     0.518231     0.283019             0.0625    0.313084  0.551020   \n",
       "\n",
       "   city_mpg  highway_mpg    price  \n",
       "0  0.222222     0.289474  13495.0  \n",
       "1  0.222222     0.289474  16500.0  \n",
       "2  0.166667     0.263158  16500.0  \n",
       "3  0.305556     0.368421  13950.0  \n",
       "4  0.138889     0.157895  17450.0  "
      ]
     },
     "execution_count": 39,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "norm_cars.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Univariate Model Building"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "In our first journey in machine learning, we will build several univariate KNN models to see which single feature best predicts the target (`price`)."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.neighbors import KNeighborsRegressor\n",
    "from sklearn.metrics import mean_squared_error"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {},
   "outputs": [],
   "source": [
    "def uni_knn_train_test(train_col, targ_col, df, k = 5):\n",
    "    '''\n",
    "    Trains and validates using KNN to predict the targ_col\n",
    "    from the train_col.\n",
    "    \n",
    "    input\n",
    "    -----\n",
    "    train_col: str\n",
    "    targ_col: str\n",
    "    df: pd.DataFrame\n",
    "    k: int\n",
    "    '''\n",
    "    # define feature and target\n",
    "    np.random.seed(1)\n",
    "    feature = [train_col]\n",
    "    target = targ_col\n",
    "    \n",
    "    # shuffle the order of dataset\n",
    "    shuffled = df.loc[np.random.permutation(df.index)].copy()\n",
    "    shuffled = shuffled.reindex()\n",
    "    \n",
    "    # divide the dataset into train and test dataframes\n",
    "    eighty = round(len(shuffled) * 0.8)\n",
    "    train_df = shuffled.loc[:eighty,:]\n",
    "    test_df = shuffled.loc[eighty:,:]\n",
    "    \n",
    "    # initiate and fit our model to the dataset\n",
    "    knn = KNeighborsRegressor(n_neighbors = k, algorithm = 'brute')\n",
    "    knn.fit(train_df[feature],train_df[target])\n",
    "    \n",
    "    # predict using our KNN algorithm\n",
    "    prediction = knn.predict(test_df[feature])\n",
    "    \n",
    "    # calculate the evaluating statistic (root mean squared error)\n",
    "    mse = mean_squared_error(test_df[target],prediction)\n",
    "    rmse = np.sqrt(mse)\n",
    "    \n",
    "    return rmse"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {
    "collapsed": false,
    "jupyter": {
     "outputs_hidden": false
    }
   },
   "outputs": [],
   "source": [
    "# get the rmse for each feature\n",
    "rmse_dic = {}\n",
    "\n",
    "for col in norm_cols:\n",
    "    rmse_dic[col] = uni_knn_train_test(col,'price',norm_cars)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {
    "collapsed": false,
    "jupyter": {
     "outputs_hidden": false
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "engine_size          2702.193221\n",
       "curb_weight          3078.149019\n",
       "city_mpg             3078.576726\n",
       "wheel_base           4006.715263\n",
       "width                4524.882412\n",
       "highway_mpg          4904.796976\n",
       "horsepower           5105.442424\n",
       "length               6575.213755\n",
       "compression_ratio    7218.477120\n",
       "height               7824.389337\n",
       "peak_rpm             7989.148097\n",
       "normalized_losses    8218.834488\n",
       "dtype: float64"
      ]
     },
     "execution_count": 43,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "k_5_uni_knn = pd.Series(rmse_dic).sort_values()\n",
    "k_5_uni_knn"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Based on our simple univariate KNN model (using k=5) so far, `engine_size` and `curb_weight` or `city_mpg` were best able to predict the `price` of a car.\n",
    "\n",
    "We will next figure out whether the hyperparameter can be tweeked for more accurate results."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {
    "collapsed": false,
    "jupyter": {
     "outputs_hidden": false
    }
   },
   "outputs": [],
   "source": [
    "rmse_dic_multi = {}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {
    "collapsed": false,
    "jupyter": {
     "outputs_hidden": false
    }
   },
   "outputs": [],
   "source": [
    "# get the rmse for each column, at different k neighbor values\n",
    "for col in norm_cols:\n",
    "    col_k = {}\n",
    "    for k in range(1,26):\n",
    "        rmse = uni_knn_train_test(col,'price',norm_cars,k=k)\n",
    "        col_k[k] = rmse\n",
    "    rmse_dic_multi[col] = col_k"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "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>k_value</th>\n",
       "      <th>normalized_losses</th>\n",
       "      <th>wheel_base</th>\n",
       "      <th>length</th>\n",
       "      <th>width</th>\n",
       "      <th>height</th>\n",
       "      <th>curb_weight</th>\n",
       "      <th>engine_size</th>\n",
       "      <th>compression_ratio</th>\n",
       "      <th>horsepower</th>\n",
       "      <th>peak_rpm</th>\n",
       "      <th>city_mpg</th>\n",
       "      <th>highway_mpg</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1</td>\n",
       "      <td>4586.752681</td>\n",
       "      <td>3694.385363</td>\n",
       "      <td>7426.252211</td>\n",
       "      <td>3993.545747</td>\n",
       "      <td>10556.603219</td>\n",
       "      <td>3500.730300</td>\n",
       "      <td>2690.466952</td>\n",
       "      <td>8452.798721</td>\n",
       "      <td>4011.642537</td>\n",
       "      <td>8618.094587</td>\n",
       "      <td>3640.693420</td>\n",
       "      <td>7342.577652</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2</td>\n",
       "      <td>4871.857027</td>\n",
       "      <td>3226.397865</td>\n",
       "      <td>7181.491842</td>\n",
       "      <td>3270.233772</td>\n",
       "      <td>8532.717920</td>\n",
       "      <td>3153.316783</td>\n",
       "      <td>2339.360425</td>\n",
       "      <td>7019.380479</td>\n",
       "      <td>4509.945982</td>\n",
       "      <td>8942.841801</td>\n",
       "      <td>3837.084239</td>\n",
       "      <td>4865.481396</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>3</td>\n",
       "      <td>7368.558972</td>\n",
       "      <td>3318.724214</td>\n",
       "      <td>7451.835311</td>\n",
       "      <td>4492.573560</td>\n",
       "      <td>8065.544094</td>\n",
       "      <td>3680.936941</td>\n",
       "      <td>2681.522527</td>\n",
       "      <td>6372.607770</td>\n",
       "      <td>5010.469933</td>\n",
       "      <td>8928.444914</td>\n",
       "      <td>3702.485051</td>\n",
       "      <td>4622.595563</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>4</td>\n",
       "      <td>9076.406715</td>\n",
       "      <td>4184.374575</td>\n",
       "      <td>7278.541556</td>\n",
       "      <td>4182.359319</td>\n",
       "      <td>8178.774726</td>\n",
       "      <td>3513.550019</td>\n",
       "      <td>2642.371090</td>\n",
       "      <td>8035.701246</td>\n",
       "      <td>4650.967954</td>\n",
       "      <td>9372.265469</td>\n",
       "      <td>2935.948643</td>\n",
       "      <td>4606.271168</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>5</td>\n",
       "      <td>8218.834488</td>\n",
       "      <td>4006.715263</td>\n",
       "      <td>6575.213755</td>\n",
       "      <td>4524.882412</td>\n",
       "      <td>7824.389337</td>\n",
       "      <td>3078.149019</td>\n",
       "      <td>2702.193221</td>\n",
       "      <td>7218.477120</td>\n",
       "      <td>5105.442424</td>\n",
       "      <td>7989.148097</td>\n",
       "      <td>3078.576726</td>\n",
       "      <td>4904.796976</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   k_value  normalized_losses   wheel_base       length        width  \\\n",
       "0        1        4586.752681  3694.385363  7426.252211  3993.545747   \n",
       "1        2        4871.857027  3226.397865  7181.491842  3270.233772   \n",
       "2        3        7368.558972  3318.724214  7451.835311  4492.573560   \n",
       "3        4        9076.406715  4184.374575  7278.541556  4182.359319   \n",
       "4        5        8218.834488  4006.715263  6575.213755  4524.882412   \n",
       "\n",
       "         height  curb_weight  engine_size  compression_ratio   horsepower  \\\n",
       "0  10556.603219  3500.730300  2690.466952        8452.798721  4011.642537   \n",
       "1   8532.717920  3153.316783  2339.360425        7019.380479  4509.945982   \n",
       "2   8065.544094  3680.936941  2681.522527        6372.607770  5010.469933   \n",
       "3   8178.774726  3513.550019  2642.371090        8035.701246  4650.967954   \n",
       "4   7824.389337  3078.149019  2702.193221        7218.477120  5105.442424   \n",
       "\n",
       "      peak_rpm     city_mpg  highway_mpg  \n",
       "0  8618.094587  3640.693420  7342.577652  \n",
       "1  8942.841801  3837.084239  4865.481396  \n",
       "2  8928.444914  3702.485051  4622.595563  \n",
       "3  9372.265469  2935.948643  4606.271168  \n",
       "4  7989.148097  3078.576726  4904.796976  "
      ]
     },
     "execution_count": 46,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "uni_multi_k = pd.DataFrame(rmse_dic_multi).reset_index()\n",
    "uni_multi_k.columns = ['k_value', 'normalized_losses', 'wheel_base', 'length', 'width', 'height',\n",
    "       'curb_weight', 'engine_size', 'compression_ratio', 'horsepower',\n",
    "       'peak_rpm', 'city_mpg', 'highway_mpg']\n",
    "uni_multi_k.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The dataset above gives the root mean squared error for each feature, at different k neighbor hyperparameters. It's hard to visualize which `k_value` setting did best in the KNN algorithm, so let's plot it and see what we get."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {
    "collapsed": false,
    "jupyter": {
     "outputs_hidden": false
    }
   },
   "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>k_value</th>\n",
       "      <th>feature</th>\n",
       "      <th>rmse</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1</td>\n",
       "      <td>normalized_losses</td>\n",
       "      <td>4586.752681</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2</td>\n",
       "      <td>normalized_losses</td>\n",
       "      <td>4871.857027</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>3</td>\n",
       "      <td>normalized_losses</td>\n",
       "      <td>7368.558972</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>4</td>\n",
       "      <td>normalized_losses</td>\n",
       "      <td>9076.406715</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>5</td>\n",
       "      <td>normalized_losses</td>\n",
       "      <td>8218.834488</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   k_value            feature         rmse\n",
       "0        1  normalized_losses  4586.752681\n",
       "1        2  normalized_losses  4871.857027\n",
       "2        3  normalized_losses  7368.558972\n",
       "3        4  normalized_losses  9076.406715\n",
       "4        5  normalized_losses  8218.834488"
      ]
     },
     "execution_count": 47,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Melt the dataset for plotting purposes\n",
    "uni_multi_k = uni_multi_k.melt(id_vars = 'k_value')\n",
    "uni_multi_k.columns = ['k_value','feature','rmse']\n",
    "uni_multi_k.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Text(0.5, 1.0, 'Root mean squared error by feature and hyperparameter')"
      ]
     },
     "execution_count": 48,
     "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.lineplot(x = 'k_value',y='rmse',hue='feature',data=uni_multi_k)\n",
    "plt.legend(bbox_to_anchor = (1,1))\n",
    "plt.title('Root mean squared error by feature and hyperparameter')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "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>k_value</th>\n",
       "      <th>feature</th>\n",
       "      <th>rmse</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>151</th>\n",
       "      <td>2</td>\n",
       "      <td>engine_size</td>\n",
       "      <td>2339.360425</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>153</th>\n",
       "      <td>4</td>\n",
       "      <td>engine_size</td>\n",
       "      <td>2642.371090</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>152</th>\n",
       "      <td>3</td>\n",
       "      <td>engine_size</td>\n",
       "      <td>2681.522527</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>150</th>\n",
       "      <td>1</td>\n",
       "      <td>engine_size</td>\n",
       "      <td>2690.466952</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>154</th>\n",
       "      <td>5</td>\n",
       "      <td>engine_size</td>\n",
       "      <td>2702.193221</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>155</th>\n",
       "      <td>6</td>\n",
       "      <td>engine_size</td>\n",
       "      <td>2893.906005</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>253</th>\n",
       "      <td>4</td>\n",
       "      <td>city_mpg</td>\n",
       "      <td>2935.948643</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>156</th>\n",
       "      <td>7</td>\n",
       "      <td>engine_size</td>\n",
       "      <td>3010.779611</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>129</th>\n",
       "      <td>5</td>\n",
       "      <td>curb_weight</td>\n",
       "      <td>3078.149019</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>254</th>\n",
       "      <td>5</td>\n",
       "      <td>city_mpg</td>\n",
       "      <td>3078.576726</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "     k_value      feature         rmse\n",
       "151        2  engine_size  2339.360425\n",
       "153        4  engine_size  2642.371090\n",
       "152        3  engine_size  2681.522527\n",
       "150        1  engine_size  2690.466952\n",
       "154        5  engine_size  2702.193221\n",
       "155        6  engine_size  2893.906005\n",
       "253        4     city_mpg  2935.948643\n",
       "156        7  engine_size  3010.779611\n",
       "129        5  curb_weight  3078.149019\n",
       "254        5     city_mpg  3078.576726"
      ]
     },
     "execution_count": 49,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "uni_multi_k.sort_values('rmse').head(10)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Overall, each feature has an optimal `k_value`. The `rmse` decreases as k increases, but will increase again after a certain point. \n",
    "\n",
    "The lowest `rmse`s are from `engine_size` and `curb_weight` or `city_mpg` at `k_value`s of 3, 5, 5 respectively.\n",
    "\n",
    "We will try a multivariate KNN model next to see if the `rmse` can be lowered further."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "metadata": {},
   "outputs": [],
   "source": [
    "results_univariate = uni_multi_k.copy()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Multivariate Model Building"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We will utilize a multivariate knn model with three types of validation:\n",
    "\n",
    "1. Multivariate KNN with train/test validation\n",
    "2. Multivariate KNN with holdout validation\n",
    "3. Multivariate KNN with k-fold validation\n",
    "\n",
    "Each model will undergo the following processing:\n",
    "\n",
    "* Shuffling of the order of the dataset\n",
    "* Diving the dataset into training and test datasets\n",
    "\n",
    "The difference lies in how the datasets are divided:\n",
    "\n",
    "* Train/test validation: training set will contain the first 80% of the data, test set the remaining 20%\n",
    "* Holdout validation: training and test sets will each contain 50% of the data, however 2 models will be created:\n",
    "    * Model 1: training and test sets run as usual\n",
    "    * Model 2: training sets reassigned to become test, and test sets reassigned to become training sets\n",
    "    * The two resulting RMSE's are averaged together, and then returned\n",
    "* K-fold validation: \n",
    "    * The dataset is divided into k-number\n",
    "    * Training set is all rows except those labeled as n, where n = 1 .... k\n",
    "    * Total number of models created: k\n",
    "    * Total number of RMSE's: k\n",
    "    * The RMSE's are averaged together, and then returned\n",
    "\n",
    "Instead of copy-pasting the function over and over again, we will build a `class` that processes, trains, tests, and then validates the dataset using the method of our choice."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "metadata": {},
   "outputs": [],
   "source": [
    "class knn_trainer():\n",
    "    def __init__(self, train_list, targ_col, df):\n",
    "        '''\n",
    "        Trains and validates using KNN to predict the targ_col\n",
    "        from the train_col.\n",
    "\n",
    "        input\n",
    "        -----\n",
    "        train_col: str\n",
    "        targ_col: str\n",
    "        df: pd.DataFrame\n",
    "        k: int\n",
    "        valid: str\n",
    "        '''\n",
    "\n",
    "        # define feature and target\n",
    "        np.random.seed(1)\n",
    "        self.feature = train_list\n",
    "        self.target = targ_col\n",
    "\n",
    "        # shuffle the order of dataset\n",
    "        shuffled = df.loc[np.random.permutation(df.index)].copy()\n",
    "        shuffled = shuffled.reindex()\n",
    "        self.shuffled = shuffled.reset_index(drop=True)\n",
    "        \n",
    "    def train_test(self, k=5):\n",
    "        '''\n",
    "        Splits the dataset 80-20, train-test\n",
    "        '''\n",
    "        self.valid = \"train_test\"\n",
    "        # divide the dataset into train and test dataframes\n",
    "        eighty = round(len(self.shuffled) * 0.8)\n",
    "        train_df = self.shuffled.loc[:eighty,:]\n",
    "        self.test_df = self.shuffled.loc[eighty:,:]\n",
    "\n",
    "        # initiate and fit our model to the dataset\n",
    "        knn = KNeighborsRegressor(n_neighbors = k, algorithm = 'brute')\n",
    "        knn.fit(train_df[self.feature],train_df[self.target])\n",
    "\n",
    "        # predict using our KNN algorithm\n",
    "        self.prediction = knn.predict(self.test_df[self.feature])\n",
    "        \n",
    "    def hold_test(self, k=5):\n",
    "        '''\n",
    "        Splits the dataset 50-50, train-test, for holdout validation\n",
    "        '''\n",
    "        self.valid = \"holdout\"\n",
    "        feature = self.feature\n",
    "        target = self.target\n",
    "        shuffled = self.shuffled\n",
    "        \n",
    "        # divide the dataset into train and test dataframes\n",
    "        fifty = round(len(shuffled) * 0.5)\n",
    "        split_one = shuffled.loc[:fifty,:].copy()\n",
    "        split_two = shuffled.loc[fifty:,:].copy()\n",
    "        \n",
    "        ## split_one ##\n",
    "        train_one = split_one\n",
    "        self.test_one = split_two\n",
    "        # initiate and fit our model\n",
    "        knn = KNeighborsRegressor(n_neighbors = k, algorithm = 'brute')\n",
    "        knn.fit(train_one[feature],train_one[target])        \n",
    "        # predict using the model\n",
    "        self.predict_one = knn.predict(self.test_one[self.feature])\n",
    "\n",
    "        ## split_two ##\n",
    "        train_two = split_two\n",
    "        self.test_two = split_one\n",
    "        # initiate and fit our model\n",
    "        knn = KNeighborsRegressor(n_neighbors = k, algorithm = 'brute')\n",
    "        knn.fit(train_two[feature],train_two[target])\n",
    "        # predict using the model\n",
    "        self.predict_two = knn.predict(self.test_two[feature])\n",
    "        \n",
    "    def kfold_test(self,k=5,fold=5):\n",
    "        '''\n",
    "        Splits the dataset k times, for k-fold validation\n",
    "        '''\n",
    "        self.valid = \"kfold\"        \n",
    "        k_df = self.shuffled\n",
    "        feature = self.feature\n",
    "        target = self.target\n",
    "        \n",
    "        # label the folds\n",
    "        split = round(len(k_df)/fold)\n",
    "        i = 0\n",
    "        for n in range(1,fold):\n",
    "            k_df.loc[i:i+split,'fold'] = n # labels n number of rows for folding\n",
    "            i += split\n",
    "        k_df.loc[i:,'fold'] = fold # finish off with the last bit\n",
    "        \n",
    "        # initiate and fit our model\n",
    "        knn = KNeighborsRegressor(n_neighbors = k, algorithm = 'brute')\n",
    "        \n",
    "        fold_predictions = {}\n",
    "        fold_tests = {}\n",
    "        for n in range(1,fold+1):\n",
    "            train_df = k_df[k_df['fold'] != n]\n",
    "            test_df = k_df[k_df['fold'] == n]\n",
    "            fold_tests[f'fold_{n}'] = test_df\n",
    "            # train fold model\n",
    "            knn.fit(train_df[feature],train_df[target])\n",
    "            # predict from fold model\n",
    "            fold_predictions[f'fold_{n}'] = knn.predict(test_df[feature])\n",
    "            \n",
    "        self.fold_predictions = fold_predictions\n",
    "        self.fold_tests = fold_tests\n",
    "        \n",
    "    def validation(self):\n",
    "        if self.valid == 'train_test':\n",
    "            # calculate the evaluating statistic (root mean squared error)\n",
    "            mse = mean_squared_error(self.test_df[self.target],self.prediction)\n",
    "            rmse = np.sqrt(mse)\n",
    "            return rmse\n",
    "        elif self.valid == 'holdout':\n",
    "            target = self.target\n",
    "            mse_one = mean_squared_error(self.test_one[target],self.predict_one)\n",
    "            rmse_one = np.sqrt(mse_one)\n",
    "            mse_two = mean_squared_error(self.test_two[target],self.predict_two)\n",
    "            rmse_two = np.sqrt(mse_two)\n",
    "\n",
    "            avg_rmse = (rmse_one + rmse_two)/2\n",
    "            return avg_rmse\n",
    "        elif self.valid == 'kfold':\n",
    "            fold_predictions = self.fold_predictions\n",
    "            fold_tests = self.fold_tests\n",
    "            \n",
    "            rmse_list = []\n",
    "            for k, v in fold_predictions.items():\n",
    "                mse = mean_squared_error(fold_tests[k]['price'],fold_predictions[k])\n",
    "                rmse = np.sqrt(mse)\n",
    "                rmse_list.append(rmse)\n",
    "                        \n",
    "            avg_rmse = np.mean(rmse_list)\n",
    "            return avg_rmse"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "In order to use a multivariate model we first need to have some idea of what features would best predict the `price`."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "metadata": {
    "collapsed": false,
    "jupyter": {
     "outputs_hidden": false
    }
   },
   "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>index</th>\n",
       "      <th>0</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>engine_size</td>\n",
       "      <td>2702.193221</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>curb_weight</td>\n",
       "      <td>3078.149019</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>city_mpg</td>\n",
       "      <td>3078.576726</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>wheel_base</td>\n",
       "      <td>4006.715263</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>width</td>\n",
       "      <td>4524.882412</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>highway_mpg</td>\n",
       "      <td>4904.796976</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>horsepower</td>\n",
       "      <td>5105.442424</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>length</td>\n",
       "      <td>6575.213755</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>compression_ratio</td>\n",
       "      <td>7218.477120</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>height</td>\n",
       "      <td>7824.389337</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>peak_rpm</td>\n",
       "      <td>7989.148097</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>normalized_losses</td>\n",
       "      <td>8218.834488</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                index            0\n",
       "0         engine_size  2702.193221\n",
       "1         curb_weight  3078.149019\n",
       "2            city_mpg  3078.576726\n",
       "3          wheel_base  4006.715263\n",
       "4               width  4524.882412\n",
       "5         highway_mpg  4904.796976\n",
       "6          horsepower  5105.442424\n",
       "7              length  6575.213755\n",
       "8   compression_ratio  7218.477120\n",
       "9              height  7824.389337\n",
       "10           peak_rpm  7989.148097\n",
       "11  normalized_losses  8218.834488"
      ]
     },
     "execution_count": 52,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "best_df = k_5_uni_knn.reset_index()\n",
    "best_df"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "According to our univariate modeling above, the top five 'best' features (using a default `k_value` of 5) for the task are:\n",
    "\n",
    "* `engine_size`\n",
    "* `curb_weight`\n",
    "* `city_mpg`\n",
    "* `wheel_base`\n",
    "* `width`"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "metadata": {
    "collapsed": false,
    "jupyter": {
     "outputs_hidden": false
    }
   },
   "outputs": [],
   "source": [
    "multi_dic = {}\n",
    "# Using all features for comparison\n",
    "features = best_df['index']\n",
    "# Initiate the knn_trainer class\n",
    "k = knn_trainer(features,'price',norm_cars)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Run the KNN test\n",
    "k.train_test()\n",
    "# Validate, return the rmse\n",
    "multi_dic[12] = k.validation()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Using best X from the univariate model\n",
    "for x in [2,3,4,5]:\n",
    "    features = list(best_df['index'][0:x])\n",
    "    k = knn_trainer(features,'price',norm_cars)\n",
    "    k.train_test()\n",
    "    multi_dic[x] = [k.validation()]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "metadata": {
    "collapsed": false,
    "jupyter": {
     "outputs_hidden": false
    }
   },
   "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>top_x_features</th>\n",
       "      <th>rmse</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>3</td>\n",
       "      <td>2695.360875</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>12</td>\n",
       "      <td>2723.746942</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2</td>\n",
       "      <td>2802.106282</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>5</td>\n",
       "      <td>2874.044709</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>4</td>\n",
       "      <td>3238.163743</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   top_x_features         rmse\n",
       "0               3  2695.360875\n",
       "1              12  2723.746942\n",
       "2               2  2802.106282\n",
       "3               5  2874.044709\n",
       "4               4  3238.163743"
      ]
     },
     "execution_count": 56,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "multi_k_five = pd.DataFrame(multi_dic).T.sort_values(0).reset_index()\n",
    "multi_k_five.columns = ['top_x_features','rmse']\n",
    "multi_k_five.sort_values('rmse')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Text(0, 0.5, 'RMSE')"
      ]
     },
     "execution_count": 57,
     "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.barplot(x='top_x_features',y='rmse',data=multi_k_five)\n",
    "plt.title('RMSE based on number of features used')\n",
    "plt.xlabel(\"Top x features\")\n",
    "plt.ylabel('RMSE')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "According to our multivariate model (with k neighbors of 5) the best `price` prediction we could get is with either 3 or 4 of the top univariate features. Using only 2 or as much as 5 features resulted in larger mean errors, and using all features did worse of all."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Hyperparameter Optimization"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "While we have an idea of which features to use, the question now is how wide to cast our net? How many k neighbors would be the most accurate way to predict the `price` of cars on the market?"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "metadata": {
    "collapsed": false,
    "jupyter": {
     "outputs_hidden": false
    }
   },
   "outputs": [],
   "source": [
    "hyper_k_rmse_dic = {}\n",
    "\n",
    "# Run multivariate KNN using the top x features, as determined when k=5 was assumed.\n",
    "for x in [3,12,2]:\n",
    "    temp_dic = {}\n",
    "    # test with k values 1-25\n",
    "    for k in range(1,26):\n",
    "        features = list(best_df['index'][0:x])\n",
    "        kn = knn_trainer(features,'price',norm_cars)\n",
    "        kn.train_test(k=k)\n",
    "        temp_dic[k] = kn.validation()\n",
    "    hyper_k_rmse_dic[\"{0}_best_features\".format(x)] = temp_dic"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 59,
   "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>k_value</th>\n",
       "      <th>3_best_features</th>\n",
       "      <th>12_best_features</th>\n",
       "      <th>2_best_features</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1</td>\n",
       "      <td>3078.374015</td>\n",
       "      <td>2387.247463</td>\n",
       "      <td>2874.096214</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2</td>\n",
       "      <td>2917.911293</td>\n",
       "      <td>2521.935891</td>\n",
       "      <td>2880.260859</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>3</td>\n",
       "      <td>2817.965489</td>\n",
       "      <td>2617.200376</td>\n",
       "      <td>2917.587428</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>4</td>\n",
       "      <td>2805.922992</td>\n",
       "      <td>2743.951650</td>\n",
       "      <td>2862.923153</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>5</td>\n",
       "      <td>2695.360875</td>\n",
       "      <td>2723.746942</td>\n",
       "      <td>2802.106282</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   k_value  3_best_features  12_best_features  2_best_features\n",
       "0        1      3078.374015       2387.247463      2874.096214\n",
       "1        2      2917.911293       2521.935891      2880.260859\n",
       "2        3      2817.965489       2617.200376      2917.587428\n",
       "3        4      2805.922992       2743.951650      2862.923153\n",
       "4        5      2695.360875       2723.746942      2802.106282"
      ]
     },
     "execution_count": 59,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "multiv_multi_k = pd.DataFrame(hyper_k_rmse_dic).reset_index()\n",
    "cols = list(multiv_multi_k.columns)\n",
    "cols.insert(0,'k_value')\n",
    "cols.remove('index')\n",
    "multiv_multi_k.columns = cols\n",
    "multiv_multi_k.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "With `k_value`s ranging from 1 through 25, it is very difficult to determine the most optimal hyperparameter settings. We will graph the above table."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "metadata": {
    "collapsed": false,
    "jupyter": {
     "outputs_hidden": false
    }
   },
   "outputs": [],
   "source": [
    "multiv_multi_k = multiv_multi_k.melt(id_vars = 'k_value')\n",
    "multiv_multi_k.columns = ['k_value','top_features','rmse']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 61,
   "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>k_value</th>\n",
       "      <th>top_features</th>\n",
       "      <th>rmse</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>25</th>\n",
       "      <td>1</td>\n",
       "      <td>12_best_features</td>\n",
       "      <td>2387.247463</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>26</th>\n",
       "      <td>2</td>\n",
       "      <td>12_best_features</td>\n",
       "      <td>2521.935891</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>27</th>\n",
       "      <td>3</td>\n",
       "      <td>12_best_features</td>\n",
       "      <td>2617.200376</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>31</th>\n",
       "      <td>7</td>\n",
       "      <td>12_best_features</td>\n",
       "      <td>2673.081599</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>5</td>\n",
       "      <td>3_best_features</td>\n",
       "      <td>2695.360875</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>56</th>\n",
       "      <td>7</td>\n",
       "      <td>2_best_features</td>\n",
       "      <td>2707.405602</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>57</th>\n",
       "      <td>8</td>\n",
       "      <td>2_best_features</td>\n",
       "      <td>2710.386120</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>55</th>\n",
       "      <td>6</td>\n",
       "      <td>2_best_features</td>\n",
       "      <td>2723.207202</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>29</th>\n",
       "      <td>5</td>\n",
       "      <td>12_best_features</td>\n",
       "      <td>2723.746942</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>28</th>\n",
       "      <td>4</td>\n",
       "      <td>12_best_features</td>\n",
       "      <td>2743.951650</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "    k_value      top_features         rmse\n",
       "25        1  12_best_features  2387.247463\n",
       "26        2  12_best_features  2521.935891\n",
       "27        3  12_best_features  2617.200376\n",
       "31        7  12_best_features  2673.081599\n",
       "4         5   3_best_features  2695.360875\n",
       "56        7   2_best_features  2707.405602\n",
       "57        8   2_best_features  2710.386120\n",
       "55        6   2_best_features  2723.207202\n",
       "29        5  12_best_features  2723.746942\n",
       "28        4  12_best_features  2743.951650"
      ]
     },
     "execution_count": 61,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "multiv_multi_k.sort_values('rmse').head(10)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 62,
   "metadata": {},
   "outputs": [],
   "source": [
    "results_multi_traintest = multiv_multi_k.copy()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 63,
   "metadata": {
    "collapsed": false,
    "jupyter": {
     "outputs_hidden": false
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Text(0.5, 1.0, 'Root mean squared error by number of multivariate features and hyperparameter')"
      ]
     },
     "execution_count": 63,
     "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.lineplot(x = 'k_value',y='rmse',hue='top_features',data=multiv_multi_k)\n",
    "plt.legend(bbox_to_anchor = (1,1))\n",
    "plt.title('Root mean squared error by number of multivariate features and hyperparameter')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Each of the previously found models have `k_value`=1 as their optimal hyperparameters. \n",
    "\n",
    "* By using the top 4 features from our univariate model, and a `k_value`=1, we brought the `rmse` down to 1948.\n",
    "* By using the top 3 features from our univariate model, and a `k_value`=1, we brought the `rmse` down to 1960.\n",
    "* By using the top 5 features from our univariate model, and a `k_value`=1, we brought the `rmse` down to 2200."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Conclusion"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The most accurate model we built with this dataset was a multivate KNN model:\n",
    "\n",
    "* `k_value` of 1\n",
    "* `algorithm` as \"brute\"\n",
    "* Using these features:\n",
    "    * `highway_mpg`\n",
    "    * `engine_size`\n",
    "    * `width`\n",
    "    * `city_mpg`\n",
    "    \n",
    "Individually these features were only able to predice the `price` with an `rmse` of around 3000-3500. Together, however, the final `rmse` value was decreased by 40% to just 1948.\n",
    "\n",
    "Based on our analysis the car market value is best predicted with those four variables."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# From Dataquest\n",
    "\n",
    "That's it for the guided steps. Here are some ideas for next steps:\n",
    "\n",
    "* Modify the `knn_train_test()` function to use k-fold cross validation instead of test/train validation.\n",
    "* Modify the `knn_train_test()` function to perform the data cleaning as well.\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# K-fold cross validation"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Holdout validation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 64,
   "metadata": {},
   "outputs": [],
   "source": [
    "# find the univariate rmse using holdout validation\n",
    "uni_rmse_hold = {}\n",
    "for col in norm_cols:\n",
    "    knn = knn_trainer([col],'price',norm_cars)\n",
    "    knn.hold_test(k=5)\n",
    "    rmse = knn.validation()\n",
    "    uni_rmse_hold[col] = [rmse]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 65,
   "metadata": {
    "comments": [
     {
      "body": [
       {
        "created": "2020-08-27T02:55:18.926Z",
        "creator": {
         "image": "https://avatars3.githubusercontent.com/u/8731022?v=4",
         "name": "pomkos",
         "user": "pomkos"
        },
        "edited": false,
        "value": "Compare these results with the univariate, train/test model."
       }
      ],
      "id": "anno/0",
      "resolved": false,
      "total": 1
     }
    ]
   },
   "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>feature</th>\n",
       "      <th>rmse</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>engine_size</td>\n",
       "      <td>3335.017797</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>horsepower</td>\n",
       "      <td>3975.628604</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>curb_weight</td>\n",
       "      <td>4121.771559</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>highway_mpg</td>\n",
       "      <td>4309.933136</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>width</td>\n",
       "      <td>4415.692641</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       feature         rmse\n",
       "0  engine_size  3335.017797\n",
       "1   horsepower  3975.628604\n",
       "2  curb_weight  4121.771559\n",
       "3  highway_mpg  4309.933136\n",
       "4        width  4415.692641"
      ]
     },
     "execution_count": 65,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "uni_rmse_hold = pd.DataFrame(uni_rmse_hold).T.reset_index()\n",
    "uni_rmse_hold.columns = ['feature','rmse']\n",
    "uni_rmse_hold = uni_rmse_hold.sort_values('rmse').reset_index(drop=True)\n",
    "uni_rmse_hold.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 66,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(array([ 0,  1,  2,  3,  4,  5,  6,  7,  8,  9, 10, 11]),\n",
       " [Text(0, 0, 'engine_size'),\n",
       "  Text(1, 0, 'horsepower'),\n",
       "  Text(2, 0, 'curb_weight'),\n",
       "  Text(3, 0, 'highway_mpg'),\n",
       "  Text(4, 0, 'width'),\n",
       "  Text(5, 0, 'city_mpg'),\n",
       "  Text(6, 0, 'length'),\n",
       "  Text(7, 0, 'wheel_base'),\n",
       "  Text(8, 0, 'compression_ratio'),\n",
       "  Text(9, 0, 'normalized_losses'),\n",
       "  Text(10, 0, 'height'),\n",
       "  Text(11, 0, 'peak_rpm')])"
      ]
     },
     "execution_count": 66,
     "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.barplot(x='feature',y='rmse',data=uni_rmse_hold)\n",
    "plt.title('RMSE vs Feature, using holdout validation')\n",
    "plt.xticks(rotation=45, horizontalalignment='right')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Univariate KNN with holdout variation resulted in the following top five features:\n",
    "\n",
    "* `engine_size`\n",
    "* `highway_mpg`\n",
    "* `width`\n",
    "* `curb_weight`\n",
    "* `city_mpg`"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 67,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Using best X from the univariate model\n",
    "multi_hold = {}\n",
    "for x in [2,3,4,5,12]:\n",
    "    feature = list(uni_rmse_hold['feature'][0:x])\n",
    "    kn = knn_trainer(feature, 'price', norm_cars)\n",
    "    kn.hold_test(k=5)\n",
    "    rmse = kn.validation()\n",
    "    multi_hold[x] = [rmse]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 68,
   "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>top_x_features</th>\n",
       "      <th>rmse</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>2</td>\n",
       "      <td>3214.100274</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>4</td>\n",
       "      <td>3315.148162</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>5</td>\n",
       "      <td>3386.339043</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>3</td>\n",
       "      <td>3401.729407</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>12</td>\n",
       "      <td>3906.435976</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   top_x_features         rmse\n",
       "0               2  3214.100274\n",
       "1               4  3315.148162\n",
       "2               5  3386.339043\n",
       "3               3  3401.729407\n",
       "4              12  3906.435976"
      ]
     },
     "execution_count": 68,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "multi_hold = pd.DataFrame(multi_hold).T.reset_index()\n",
    "multi_hold.columns = ['top_x_features','rmse']\n",
    "multi_hold = multi_hold.sort_values('rmse').reset_index(drop=True)\n",
    "multi_hold.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 69,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Text(0.5, 1.0, 'RMSE vs Top X Features, using holdout validation')"
      ]
     },
     "execution_count": 69,
     "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.barplot(x='top_x_features',y='rmse',data=multi_hold)\n",
    "plt.title('RMSE vs Top X Features, using holdout validation')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Each multivariate KNN performed similarly with `k_value`=5 as calculated by holdout validation, except when using all 12 features."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 70,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Lets tune the hyperparameters for KNN holdout variation\n",
    "top_k_hold = {}\n",
    "for x in [2,4,5]:\n",
    "    k_dic = {}\n",
    "    for k in range(1,26):\n",
    "        feature = list(uni_rmse_hold['feature'][0:x])\n",
    "        kn = knn_trainer(feature, 'price', norm_cars)\n",
    "        kn.hold_test(k=k)\n",
    "        rmse = kn.validation()\n",
    "        k_dic[k] = rmse\n",
    "    top_k_hold[f'top_{x}_features'] = k_dic"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 71,
   "metadata": {},
   "outputs": [],
   "source": [
    "top_k_hold = pd.DataFrame(top_k_hold)\n",
    "top_k_hold = top_k_hold.reset_index()\n",
    "top_k_hold = top_k_hold.melt(id_vars = 'index')\n",
    "top_k_hold.columns = ['k_value','features','rmse']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 72,
   "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>k_value</th>\n",
       "      <th>features</th>\n",
       "      <th>rmse</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>51</th>\n",
       "      <td>2</td>\n",
       "      <td>top_5_features</td>\n",
       "      <td>2892.335953</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>3</td>\n",
       "      <td>top_2_features</td>\n",
       "      <td>2897.000178</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>26</th>\n",
       "      <td>2</td>\n",
       "      <td>top_4_features</td>\n",
       "      <td>3000.489562</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25</th>\n",
       "      <td>1</td>\n",
       "      <td>top_4_features</td>\n",
       "      <td>3004.732820</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>52</th>\n",
       "      <td>3</td>\n",
       "      <td>top_5_features</td>\n",
       "      <td>3048.822235</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "    k_value        features         rmse\n",
       "51        2  top_5_features  2892.335953\n",
       "2         3  top_2_features  2897.000178\n",
       "26        2  top_4_features  3000.489562\n",
       "25        1  top_4_features  3004.732820\n",
       "52        3  top_5_features  3048.822235"
      ]
     },
     "execution_count": 72,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "top_k_hold.sort_values('rmse').head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 73,
   "metadata": {},
   "outputs": [],
   "source": [
    "results_multi_holdout = top_k_hold.copy()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 74,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Text(0.5, 0, 'K value')"
      ]
     },
     "execution_count": 74,
     "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.lineplot(x = 'k_value', y='rmse', hue='features', data=top_k_hold)\n",
    "plt.legend(bbox_to_anchor = (1,1))\n",
    "plt.title('Root mean squared error by number of multivariate kfold features and hyperparameter')\n",
    "plt.ylabel('RMSE')\n",
    "plt.xlabel('K value')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Very similar results as with the train/test validation. Using the`top_4_features` with `k_value`=1 resulted in the smallest `rmse` for the dataset."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# k-fold validation"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Using both manual folds and the `KFold` and `cross_val_score` functions from the `sklearn.model_selection` library."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 75,
   "metadata": {},
   "outputs": [],
   "source": [
    "uni_rmse_fold = {}\n",
    "\n",
    "for col in norm_cols:\n",
    "    kn = knn_trainer([col], 'price', norm_cars)\n",
    "    kn.kfold_test(k=5,fold=5)\n",
    "    rmse = kn.validation()\n",
    "    uni_rmse_fold[col] = [rmse]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 76,
   "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>feature</th>\n",
       "      <th>rmse</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>engine_size</td>\n",
       "      <td>3062.585799</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>horsepower</td>\n",
       "      <td>3796.442472</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>width</td>\n",
       "      <td>4134.792903</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>highway_mpg</td>\n",
       "      <td>4201.534931</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>curb_weight</td>\n",
       "      <td>4332.560251</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       feature         rmse\n",
       "0  engine_size  3062.585799\n",
       "1   horsepower  3796.442472\n",
       "2        width  4134.792903\n",
       "3  highway_mpg  4201.534931\n",
       "4  curb_weight  4332.560251"
      ]
     },
     "execution_count": 76,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "uni_rmse_fold = pd.DataFrame(uni_rmse_fold).T.reset_index()\n",
    "uni_rmse_fold.columns = ['feature','rmse']\n",
    "uni_rmse_fold = uni_rmse_fold.sort_values('rmse').reset_index(drop=True)\n",
    "uni_rmse_fold.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 77,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(array([ 0,  1,  2,  3,  4,  5,  6,  7,  8,  9, 10, 11]),\n",
       " [Text(0, 0, 'engine_size'),\n",
       "  Text(1, 0, 'horsepower'),\n",
       "  Text(2, 0, 'width'),\n",
       "  Text(3, 0, 'highway_mpg'),\n",
       "  Text(4, 0, 'curb_weight'),\n",
       "  Text(5, 0, 'city_mpg'),\n",
       "  Text(6, 0, 'length'),\n",
       "  Text(7, 0, 'normalized_losses'),\n",
       "  Text(8, 0, 'wheel_base'),\n",
       "  Text(9, 0, 'compression_ratio'),\n",
       "  Text(10, 0, 'peak_rpm'),\n",
       "  Text(11, 0, 'height')])"
      ]
     },
     "execution_count": 77,
     "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.barplot(x='feature',y='rmse',data=uni_rmse_fold)\n",
    "plt.title('RMSE vs Feature, using k-fold validation')\n",
    "plt.xticks(rotation=45, horizontalalignment='right')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Using univariate KNN with k-fold validation (`fold=5`) the top six features were:\n",
    "\n",
    "* `engine_size`\n",
    "* `horsepower`\n",
    "* `width`\n",
    "* `highway_mpg`\n",
    "* `curb_weight`\n",
    "* `city_mpg`\n",
    "\n",
    "Let's see how our model does with multiple features."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 78,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Using best X from the univariate model\n",
    "multi_fold = {}\n",
    "for x in [2,3,4,5,6,12]:\n",
    "    feature = list(uni_rmse_fold['feature'][0:x])\n",
    "    kn = knn_trainer(feature, 'price', norm_cars)\n",
    "    kn.kfold_test(k=5)\n",
    "    rmse = kn.validation()\n",
    "    multi_fold[x] = [rmse]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 79,
   "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>top_x_features</th>\n",
       "      <th>rmse</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>2</td>\n",
       "      <td>2925.897105</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>3</td>\n",
       "      <td>3058.214925</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>5</td>\n",
       "      <td>3111.035110</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>4</td>\n",
       "      <td>3113.754491</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>6</td>\n",
       "      <td>3123.471067</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   top_x_features         rmse\n",
       "0               2  2925.897105\n",
       "1               3  3058.214925\n",
       "2               5  3111.035110\n",
       "3               4  3113.754491\n",
       "4               6  3123.471067"
      ]
     },
     "execution_count": 79,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "multi_fold = pd.DataFrame(multi_fold).T.reset_index()\n",
    "multi_fold.columns = ['top_x_features','rmse']\n",
    "multi_fold = multi_fold.sort_values('rmse').reset_index(drop=True)\n",
    "multi_fold.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 80,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Text(0.5, 1.0, 'RMSE vs Top X Features, using kfold validation')"
      ]
     },
     "execution_count": 80,
     "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.barplot(x='top_x_features',y='rmse',data=multi_fold)\n",
    "plt.title('RMSE vs Top X Features, using kfold validation')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Each multivariate KNN performed similarly with `k_value`=5 as calculated by kfold (`fold=5`) validation, except when using all 12 features."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 81,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Lets tune the hyperparameters for KNN kfold variation\n",
    "top_k_fold = {}\n",
    "for x in [2,3,4,5,12]:\n",
    "    k_dic = {}\n",
    "    for k in range(1,26):\n",
    "        feature = list(uni_rmse_fold['feature'][0:x])\n",
    "        kn = knn_trainer(feature, 'price', norm_cars)\n",
    "        kn.kfold_test(k=k)\n",
    "        rmse = kn.validation()\n",
    "        k_dic[k] = rmse\n",
    "    top_k_fold[f'top_{x}_features'] = k_dic"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 82,
   "metadata": {},
   "outputs": [],
   "source": [
    "top_k_fold = pd.DataFrame(top_k_fold)\n",
    "top_k_fold = top_k_fold.reset_index()\n",
    "top_k_fold = top_k_fold.melt(id_vars = 'index')\n",
    "top_k_fold.columns = ['k_value','features','rmse']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 83,
   "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>k_value</th>\n",
       "      <th>features</th>\n",
       "      <th>rmse</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>75</th>\n",
       "      <td>1</td>\n",
       "      <td>top_5_features</td>\n",
       "      <td>2464.852485</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>100</th>\n",
       "      <td>1</td>\n",
       "      <td>top_12_features</td>\n",
       "      <td>2723.191172</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>76</th>\n",
       "      <td>2</td>\n",
       "      <td>top_5_features</td>\n",
       "      <td>2727.497091</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>26</th>\n",
       "      <td>2</td>\n",
       "      <td>top_3_features</td>\n",
       "      <td>2752.590633</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>101</th>\n",
       "      <td>2</td>\n",
       "      <td>top_12_features</td>\n",
       "      <td>2763.931554</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>3</td>\n",
       "      <td>top_2_features</td>\n",
       "      <td>2771.541399</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>27</th>\n",
       "      <td>3</td>\n",
       "      <td>top_3_features</td>\n",
       "      <td>2772.509693</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>77</th>\n",
       "      <td>3</td>\n",
       "      <td>top_5_features</td>\n",
       "      <td>2784.213051</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>52</th>\n",
       "      <td>3</td>\n",
       "      <td>top_4_features</td>\n",
       "      <td>2804.001210</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>51</th>\n",
       "      <td>2</td>\n",
       "      <td>top_4_features</td>\n",
       "      <td>2826.523842</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "     k_value         features         rmse\n",
       "75         1   top_5_features  2464.852485\n",
       "100        1  top_12_features  2723.191172\n",
       "76         2   top_5_features  2727.497091\n",
       "26         2   top_3_features  2752.590633\n",
       "101        2  top_12_features  2763.931554\n",
       "2          3   top_2_features  2771.541399\n",
       "27         3   top_3_features  2772.509693\n",
       "77         3   top_5_features  2784.213051\n",
       "52         3   top_4_features  2804.001210\n",
       "51         2   top_4_features  2826.523842"
      ]
     },
     "execution_count": 83,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "top_k_fold.sort_values('rmse').head(10)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 84,
   "metadata": {},
   "outputs": [],
   "source": [
    "results_multi_kfold = top_k_fold.copy()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 85,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Text(0.5, 0, 'K value')"
      ]
     },
     "execution_count": 85,
     "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.lineplot(x = 'k_value', y='rmse', hue='features', data=top_k_fold)\n",
    "plt.legend(bbox_to_anchor = (1,1))\n",
    "plt.title('Root mean squared error by top multivariate kfold features and hyperparameter')\n",
    "plt.ylabel('RMSE')\n",
    "plt.xlabel('K value')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The best `k_value`s were on the lower end of the scale when using k-fold (`fold=5`) validation. Our best performing model used `k_value = 1`, `top_5_features`, and resulted in an `rmse = 2465`. When using k-fold validation we recommend using the following features with this dataset:\n",
    "\n",
    "* `engine_size`\n",
    "* `horsepower`\n",
    "* `width`\n",
    "* `highway_mpg`\n",
    "* `curb_weight`"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Comparison\n",
    "\n",
    "So which model-validation combination did the best?"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 86,
   "metadata": {},
   "outputs": [],
   "source": [
    "results_list = {'univariate':results_univariate,\n",
    "                'multi_traintest':results_multi_traintest,\n",
    "                'multi_holdout':results_multi_holdout,\n",
    "                'multi_kfold':results_multi_kfold}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 87,
   "metadata": {},
   "outputs": [],
   "source": [
    "results = pd.DataFrame(columns=['k_value','feature','rmse','type'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 88,
   "metadata": {},
   "outputs": [],
   "source": [
    "# format each dataframe, label it, and then append it to a big dataframe\n",
    "for k,v in results_list.items():\n",
    "    v.columns = ['k_value','feature','rmse']\n",
    "    v['type'] = k\n",
    "    results = results.append(v)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 89,
   "metadata": {},
   "outputs": [],
   "source": [
    "results_hd = results.sort_values('rmse').head(10)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 90,
   "metadata": {},
   "outputs": [],
   "source": [
    "results_hd = results_hd.astype({\n",
    "    'k_value':str,\n",
    "    'feature':str,\n",
    "    'type':str\n",
    "})"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 91,
   "metadata": {},
   "outputs": [],
   "source": [
    "results_hd['id'] = results_hd['type'] + '_' + results_hd['feature'] + '_k' + results_hd['k_value']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 92,
   "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>k_value</th>\n",
       "      <th>feature</th>\n",
       "      <th>rmse</th>\n",
       "      <th>type</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1</td>\n",
       "      <td>normalized_losses</td>\n",
       "      <td>4586.752681</td>\n",
       "      <td>univariate</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2</td>\n",
       "      <td>normalized_losses</td>\n",
       "      <td>4871.857027</td>\n",
       "      <td>univariate</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>3</td>\n",
       "      <td>normalized_losses</td>\n",
       "      <td>7368.558972</td>\n",
       "      <td>univariate</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>4</td>\n",
       "      <td>normalized_losses</td>\n",
       "      <td>9076.406715</td>\n",
       "      <td>univariate</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>5</td>\n",
       "      <td>normalized_losses</td>\n",
       "      <td>8218.834488</td>\n",
       "      <td>univariate</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "  k_value            feature         rmse        type\n",
       "0       1  normalized_losses  4586.752681  univariate\n",
       "1       2  normalized_losses  4871.857027  univariate\n",
       "2       3  normalized_losses  7368.558972  univariate\n",
       "3       4  normalized_losses  9076.406715  univariate\n",
       "4       5  normalized_losses  8218.834488  univariate"
      ]
     },
     "execution_count": 92,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "results.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 99,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<seaborn.axisgrid.FacetGrid at 0x7f0eafe7e748>"
      ]
     },
     "execution_count": 99,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 864x216 with 4 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "g = sns.FacetGrid(data=results, col='type',hue='feature', sharey=False)\n",
    "g.map(sns.lineplot,'k_value','rmse')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "There is a very wide range in `rmse`, let's zoom in on the data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 100,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<seaborn.axisgrid.FacetGrid at 0x7f0eafc31668>"
      ]
     },
     "execution_count": 100,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 864x216 with 4 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "data = results[results['rmse'] <= 4500]\n",
    "g = sns.FacetGrid(data=data, col='type',hue='feature')\n",
    "g.map(sns.lineplot,'k_value','rmse')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 109,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.legend.Legend at 0x7f0eaf476320>"
      ]
     },
     "execution_count": 109,
     "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": [
    "data = results[results['rmse'] <= 2600]\n",
    "sns.barplot(data=data, x='type',y='rmse',hue='k_value')\n",
    "plt.legend(bbox_to_anchor=(1,1))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Data Python Env",
   "language": "python",
   "name": "data_env"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.10"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}