example_predict_miles_per_gallon.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 0. Initialize common libraries"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import pandas as pd\n",
    "import seaborn as sns\n",
    "\n",
    "import matplotlib.pyplot as plt"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 1. Loading and inspecting data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Load example dataset (same as Bethany Baumann used for data visualization \n",
    "# https://notebooks.ai/Beth1126/pda-python-dataframes-and-graphs-webinar-2020-54351877)\n",
    "data = sns.load_dataset('mpg')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "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>mpg</th>\n",
       "      <th>cylinders</th>\n",
       "      <th>displacement</th>\n",
       "      <th>horsepower</th>\n",
       "      <th>weight</th>\n",
       "      <th>acceleration</th>\n",
       "      <th>model_year</th>\n",
       "      <th>origin</th>\n",
       "      <th>name</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>18.0</td>\n",
       "      <td>8</td>\n",
       "      <td>307.0</td>\n",
       "      <td>130.0</td>\n",
       "      <td>3504</td>\n",
       "      <td>12.0</td>\n",
       "      <td>70</td>\n",
       "      <td>usa</td>\n",
       "      <td>chevrolet chevelle malibu</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>15.0</td>\n",
       "      <td>8</td>\n",
       "      <td>350.0</td>\n",
       "      <td>165.0</td>\n",
       "      <td>3693</td>\n",
       "      <td>11.5</td>\n",
       "      <td>70</td>\n",
       "      <td>usa</td>\n",
       "      <td>buick skylark 320</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>18.0</td>\n",
       "      <td>8</td>\n",
       "      <td>318.0</td>\n",
       "      <td>150.0</td>\n",
       "      <td>3436</td>\n",
       "      <td>11.0</td>\n",
       "      <td>70</td>\n",
       "      <td>usa</td>\n",
       "      <td>plymouth satellite</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>16.0</td>\n",
       "      <td>8</td>\n",
       "      <td>304.0</td>\n",
       "      <td>150.0</td>\n",
       "      <td>3433</td>\n",
       "      <td>12.0</td>\n",
       "      <td>70</td>\n",
       "      <td>usa</td>\n",
       "      <td>amc rebel sst</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>17.0</td>\n",
       "      <td>8</td>\n",
       "      <td>302.0</td>\n",
       "      <td>140.0</td>\n",
       "      <td>3449</td>\n",
       "      <td>10.5</td>\n",
       "      <td>70</td>\n",
       "      <td>usa</td>\n",
       "      <td>ford torino</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "    mpg  cylinders  displacement  horsepower  weight  acceleration  \\\n",
       "0  18.0          8         307.0       130.0    3504          12.0   \n",
       "1  15.0          8         350.0       165.0    3693          11.5   \n",
       "2  18.0          8         318.0       150.0    3436          11.0   \n",
       "3  16.0          8         304.0       150.0    3433          12.0   \n",
       "4  17.0          8         302.0       140.0    3449          10.5   \n",
       "\n",
       "   model_year origin                       name  \n",
       "0          70    usa  chevrolet chevelle malibu  \n",
       "1          70    usa          buick skylark 320  \n",
       "2          70    usa         plymouth satellite  \n",
       "3          70    usa              amc rebel sst  \n",
       "4          70    usa                ford torino  "
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data.head() # dataset contains information on cars"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 2. Data cleaning"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 2.1. Column to predict"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Let us find the columns, which we want to predict and format it\n",
    "feature_to_predict = 'mpg'  # miles per gallon\n",
    "readout = data[[feature_to_predict]].astype(float)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 2.2. Features"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 2.2.1. Numeric features"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "ordinal_features = [\n",
    "    'cylinders',\n",
    "    'displacement',\n",
    "    'horsepower',\n",
    "    'weight',\n",
    "    'acceleration',\n",
    "    'model_year'\n",
    "]\n",
    "\n",
    "features_ordinal = data[ordinal_features].astype(float)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "At least some features have absent values\n"
     ]
    }
   ],
   "source": [
    "if any(features_ordinal.isnull()):\n",
    "    print('At least some features have absent values')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Make educated guesses on missing values (impute)\n",
    "from sklearn.impute import SimpleImputer   # note that sklearn will soon get better ones\n",
    "imputer = SimpleImputer(missing_values=np.nan, strategy='median')\n",
    "imputed_values = imputer.fit_transform(features_ordinal.values)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "# For consistency, store data in original format (DataFrame)\n",
    "features_ordinal = pd.DataFrame(\n",
    "    index=features_ordinal.index,\n",
    "    data=imputed_values,\n",
    "    columns=features_ordinal.columns\n",
    ")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 2.2.2. Nominal  features"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "nominal_features = [\n",
    "    'origin'\n",
    "]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.preprocessing import OneHotEncoder \n",
    "onehotencoder = OneHotEncoder() \n",
    "\n",
    "one_hot_feature = onehotencoder.fit_transform(\n",
    "    data[nominal_features]).toarray()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[0., 0., 1.],\n",
       "       [0., 0., 1.],\n",
       "       [0., 0., 1.],\n",
       "       ...,\n",
       "       [0., 0., 1.],\n",
       "       [0., 0., 1.],\n",
       "       [0., 0., 1.]])"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "one_hot_feature"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[array(['europe', 'japan', 'usa'], dtype=object)]"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "onehotencoder.categories_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "features_nominal = pd.DataFrame(\n",
    "    index=data.index, \n",
    "    columns=onehotencoder.categories_,\n",
    "    data=one_hot_feature\n",
    "\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "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 tr th {\n",
       "        text-align: left;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr>\n",
       "      <th></th>\n",
       "      <th>europe</th>\n",
       "      <th>japan</th>\n",
       "      <th>usa</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "  europe japan  usa\n",
       "0    0.0   0.0  1.0\n",
       "1    0.0   0.0  1.0\n",
       "2    0.0   0.0  1.0"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "features_nominal.head(3)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 2.2.3. Combine ordinal and nominal features"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "features = pd.concat(\n",
    "    [features_ordinal, features_nominal],\n",
    "    axis='columns'\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "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>cylinders</th>\n",
       "      <th>displacement</th>\n",
       "      <th>horsepower</th>\n",
       "      <th>weight</th>\n",
       "      <th>acceleration</th>\n",
       "      <th>model_year</th>\n",
       "      <th>(europe,)</th>\n",
       "      <th>(japan,)</th>\n",
       "      <th>(usa,)</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>8.0</td>\n",
       "      <td>307.0</td>\n",
       "      <td>130.0</td>\n",
       "      <td>3504.0</td>\n",
       "      <td>12.0</td>\n",
       "      <td>70.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>8.0</td>\n",
       "      <td>350.0</td>\n",
       "      <td>165.0</td>\n",
       "      <td>3693.0</td>\n",
       "      <td>11.5</td>\n",
       "      <td>70.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>8.0</td>\n",
       "      <td>318.0</td>\n",
       "      <td>150.0</td>\n",
       "      <td>3436.0</td>\n",
       "      <td>11.0</td>\n",
       "      <td>70.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>8.0</td>\n",
       "      <td>304.0</td>\n",
       "      <td>150.0</td>\n",
       "      <td>3433.0</td>\n",
       "      <td>12.0</td>\n",
       "      <td>70.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>8.0</td>\n",
       "      <td>302.0</td>\n",
       "      <td>140.0</td>\n",
       "      <td>3449.0</td>\n",
       "      <td>10.5</td>\n",
       "      <td>70.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   cylinders  displacement  horsepower  weight  acceleration  model_year  \\\n",
       "0        8.0         307.0       130.0  3504.0          12.0        70.0   \n",
       "1        8.0         350.0       165.0  3693.0          11.5        70.0   \n",
       "2        8.0         318.0       150.0  3436.0          11.0        70.0   \n",
       "3        8.0         304.0       150.0  3433.0          12.0        70.0   \n",
       "4        8.0         302.0       140.0  3449.0          10.5        70.0   \n",
       "\n",
       "   (europe,)  (japan,)  (usa,)  \n",
       "0        0.0       0.0     1.0  \n",
       "1        0.0       0.0     1.0  \n",
       "2        0.0       0.0     1.0  \n",
       "3        0.0       0.0     1.0  \n",
       "4        0.0       0.0     1.0  "
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "features.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 3. Machine Learning"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 3.1. Split training and testing set"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.model_selection import train_test_split"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [],
   "source": [
    "# per informal convention we will use\n",
    "#  X to designate our features and\n",
    "#  y as the value that should be predicted\n",
    "X_train, X_test, y_train, y_test = train_test_split(\n",
    "    features, \n",
    "    readout, \n",
    "    test_size=0.33)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 3.2. Initialize a Gradient Boosting Model\n",
    "Gradient Boosting will essentially perform two steps: First it will create a decision based model on the data. Second it will create a model of its own error. Together this means that 1) gradient boosting models can accomodate very distinct non-linear relationships between features and the entity to be predicted; 2) they are quite robust against being supplied too many (non-informative) features"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/tstoeger/anaconda3/lib/python3.7/site-packages/sklearn/utils/validation.py:761: DataConversionWarning: A column-vector y was passed when a 1d array was expected. Please change the shape of y to (n_samples, ), for example using ravel().\n",
      "  y = column_or_1d(y, warn=True)\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "GradientBoostingRegressor(alpha=0.9, criterion='friedman_mse', init=None,\n",
       "             learning_rate=0.1, loss='ls', max_depth=3, max_features=None,\n",
       "             max_leaf_nodes=None, min_impurity_decrease=0.0,\n",
       "             min_impurity_split=None, min_samples_leaf=1,\n",
       "             min_samples_split=2, min_weight_fraction_leaf=0.0,\n",
       "             n_estimators=100, n_iter_no_change=None, presort='auto',\n",
       "             random_state=None, subsample=1.0, tol=0.0001,\n",
       "             validation_fraction=0.1, verbose=0, warm_start=False)"
      ]
     },
     "execution_count": 19,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from sklearn import ensemble\n",
    "gradient_booster = ensemble.GradientBoostingRegressor()\n",
    "gradient_booster.fit(X_train, y_train) # train with a subset of our data"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 3.3. Test performance of our model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [],
   "source": [
    "y_predicted = gradient_booster.predict(X_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Text(0, 0.5, 'Predicted miles per gallon')"
      ]
     },
     "execution_count": 21,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 360x360 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(5,5))\n",
    "plt.scatter(y_test, y_predicted)\n",
    "plt.xlabel('Observed miles per gallon')\n",
    "plt.ylabel('Predicted miles per gallon')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [],
   "source": [
    "from scipy.stats import spearmanr"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "SpearmanrResult(correlation=0.9559594972106207, pvalue=4.489234746558822e-71)"
      ]
     },
     "execution_count": 23,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "spearmanr(y_test, y_predicted)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 3.4. Inspect models\n",
    "To understand, which features drive the prediction."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([0.00541008, 0.3792603 , 0.10535436, 0.29445748, 0.01638485,\n",
       "       0.19139826, 0.00270457, 0.0011627 , 0.00386741])"
      ]
     },
     "execution_count": 24,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "gradient_booster.feature_importances_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "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>cylinders</th>\n",
       "      <th>displacement</th>\n",
       "      <th>horsepower</th>\n",
       "      <th>weight</th>\n",
       "      <th>acceleration</th>\n",
       "      <th>model_year</th>\n",
       "      <th>(europe,)</th>\n",
       "      <th>(japan,)</th>\n",
       "      <th>(usa,)</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>8.0</td>\n",
       "      <td>307.0</td>\n",
       "      <td>130.0</td>\n",
       "      <td>3504.0</td>\n",
       "      <td>12.0</td>\n",
       "      <td>70.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>8.0</td>\n",
       "      <td>350.0</td>\n",
       "      <td>165.0</td>\n",
       "      <td>3693.0</td>\n",
       "      <td>11.5</td>\n",
       "      <td>70.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   cylinders  displacement  horsepower  weight  acceleration  model_year  \\\n",
       "0        8.0         307.0       130.0  3504.0          12.0        70.0   \n",
       "1        8.0         350.0       165.0  3693.0          11.5        70.0   \n",
       "\n",
       "   (europe,)  (japan,)  (usa,)  \n",
       "0        0.0       0.0     1.0  \n",
       "1        0.0       0.0     1.0  "
      ]
     },
     "execution_count": 25,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "features.head(2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "displacement    0.379260\n",
       "weight          0.294457\n",
       "model_year      0.191398\n",
       "horsepower      0.105354\n",
       "acceleration    0.016385\n",
       "cylinders       0.005410\n",
       "(usa,)          0.003867\n",
       "(europe,)       0.002705\n",
       "(japan,)        0.001163\n",
       "dtype: float64"
      ]
     },
     "execution_count": 26,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pd.Series(\n",
    "    index=features.columns, \n",
    "    data=gradient_booster.feature_importances_).sort_values(ascending=False)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# (optional) 4. Iterate to test robustness\n",
    "if running above multiple times, you will notice that some of the numbers (correlation between predicted and measured; and features importances) will change. This is largely because you will have different training and test sets. "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 4.1. Run multiple iterations, and gather statistics"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [],
   "source": [
    "iterations = 300"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [],
   "source": [
    "correlations = []\n",
    "\n",
    "feature_importances = pd.DataFrame(\n",
    "    index=features.columns,\n",
    "    columns=range(iterations),\n",
    "    data=np.nan\n",
    "\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [],
   "source": [
    "for j in range(iterations):\n",
    "\n",
    "    X_train, X_test, y_train, y_test = train_test_split(\n",
    "        features, readout, test_size=0.33)\n",
    "    \n",
    "    gradient_booster.fit(X_train, y_train.values.ravel())\n",
    "    y_predicted = gradient_booster.predict(X_test)\n",
    "\n",
    "    correlations.append(spearmanr(y_test, y_predicted)[0])\n",
    "    feature_importances.loc[:, j] = gradient_booster.feature_importances_"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 4.2. Inspect"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x7f85a8fa77f0>"
      ]
     },
     "execution_count": 30,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXUAAAD8CAYAAACINTRsAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDMuMC4zLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvnQurowAAIABJREFUeJzt3Xl8XNV99/HPb0ajbbSPdtmyZMnCKzZGXrDBBoMTlhAgpNCQhZCkThPSJmnap0mbpk36pE/aPlm6hKQkECiBQhIgQGISDHjBxtjIu2xhSZZXWfu+jjQzp39oDI6R0ViamTsz+r1fr3l5ljua71zP/HR07rnniDEGpZRSscFmdQCllFLBo0VdKaViiBZ1pZSKIVrUlVIqhmhRV0qpGKJFXSmlYogWdaWUiiFa1JVSKoZoUVdKqRgSF84Xy87ONiUlJeF8SaWUinp79uxpN8bkBLJtWIt6SUkJVVVV4XxJpZSKeiJyMtBttftFKaViiBZ1pZSKIVrUlVIqhmhRV0qpGKJFXSmlYogWdaWUiiFa1JVSKoZoUVdKqRiiRV0ppWJIWM8oVUqFxhO7Tk3qefesKA5yEmW1CVvqIpIoIrtF5ICIHBaRb/rvf0REjovIfv9lSejjKqWUei+BtNTdwDpjTL+IOIDtIvKi/7G/Msb8KnTxlFJKXYoJi7oxxgD9/psO/8WEMpRSSqnJCahPXUTswB6gHPihMWaXiHwO+LaIfAN4BfiqMcYduqhKxb7J9o0rdU5Ao1+MMV5jzBJgBrBcRBYCXwPmAsuALOCvx3uuiGwQkSoRqWprawtSbKWUUuO5pCGNxphuYAtwozGmyYxxAz8Dll/kOQ8aYyqNMZU5OQHN8a6UUmqSAhn9kiMiGf7rScANwFsiUuC/T4DbgepQBlVKKTWxQPrUC4BH/f3qNuAXxpjfiMirIpIDCLAf+NMQ5lRKKRWAQEa/HASuGOf+dSFJpFSM0IOeygo6TYBSSsUQLepKKRVDdO4XpaKc12cY8fgY8fpIjrfjsGtbbTrToq5UFBlwezjePsDJjgFaet209g3TO+z5g20SHTZmZCQzrzCNBQVppCU5LEqrrKBFXakI1zM0yqEz3Rxq7OF01xAADruQl5ZIeW4KGcnxJDnsxNmFwREvvUOjNLQN8MKBs2w82MTq8myum5tDQpzd4neiwkGLulIR6mz3ENvr2zl4phufgYL0RG6Yl0tZTgpFmUnE2d67m6W1b5htte1sq2vjwJlu7qqcSWm2M0zplVW0qCsVYQbdHjZWN7P3VBfxcTaumu1iRamL7NSES/o5uamJfPjKGSwryeTpvWf42Y7jfHTFLC7LTw1RchUJtKgrFUGOnO3hmX2NDI96WVuRw5o5OSTFT63bZJbLyYY1ZTyy4zg/f+Mkdy+bycKi9CAlVpFGi7pSEeL1Y+389mAThRlJ3Ll0BvnpiUH72SkJcXz66tk8uvMEv6g6jSslnoL0pKD9fBU5dOyTUhYzxvC76iZ+c7CJeQVpbFgzO6gF/ZykeDsfWzmL5Hg7T+w6xfCoN+ivoaynRV0pi+2ob2dbXTsrSrO4Z0VxSMeZpyTEcfeyYjoHRnh2XyNja+CoWKJFXSkL1bX08WJ1MwsK07h1cSE2kZC/Zmm2k/Xz8zjU2MPGQ80hfz0VXlrUlbJIR7+b/3nzFHlpY6NUwlHQz1lTkUN+WiL/tLFGu2FijBZ1pSxgjOHZfY0AfGzlrLCfGGQT4ZbLC2jsHuKnrzWE9bVVaGlRV8oCB85009A+wPsX5JPljLckQ1lOCu+bn8cDW47R0jtsSQYVfFrUlQqzoREvGw81MyMziWUlWZZm+dtb5uHxGr6/qdbSHCp4tKgrFWYv17Qw4PZw2+KisPajj2eWy8ldy2bwzN5Gba3HCC3qSoVR9+AIu493sqwki6LMyDj5Z8M1ZXh8Ph7ecdzqKCoItKgrFUav1bdjMFx7WY7VUd5W7Erm5kUFPPHGKXqHR62Oo6ZIi7pSYdLv9lB1opMlMzPJSLbm4OjF/OnaMvrcHh5/Q9dVjXYTFnURSRSR3SJyQEQOi8g3/feXisguEakTkadEJLI+pUpFmNfr2/F4DWsqsq2O8i4Li9K5Zk42D+84zojHZ3UcNQWBtNTdwDpjzGJgCXCjiKwE/hn4vjFmDtAFfDp0MZWKbsOjXnY2dLCgMI3c1ODP6xIMn1pdSlufm1dqWqyOoqZgwqJuxvT7bzr8FwOsA37lv/9R4PaQJFQqBuw73Y3b42NNReT0pV9oTUUOBemJPPnmaaujqCkIqE9dROwish9oBTYBx4BuY8y5xRHPAEWhiahU9Ks60UlheiIzMpOtjnJRdpvwR5Uz2VbXxpmuQavjqEkKqKgbY7zGmCXADGA5MG+8zcZ7rohsEJEqEalqa2ubfFKlolRj9xBNPcNcafGJRoG4q3IGAL+sOmNxEjVZlzT6xRjTDWwBVgIZInJukY0ZwNmLPOdBY0ylMaYyJydy//RUKlSqTnQSZxOWzMiwOsqEZmQmc82cHH5ZdRqvT6fljUaBjH7JEZEM//Uk4AagBtgMfNi/2b3Ac6EKqVS0GvH4OHCmm4VF6VNeli5c/njZTM72DLOtTv+yjkaBtNQLgM0ichB4E9hkjPkN8NfAX4hIPeACHgpdTKWi0+GzPQyP+qiclWl1lIDdMC+P9CQHz+8f949vFeEmXKPUGHMQuGKc+xsY619XSl3EvtPdZDnjKc12Wh0lYPFxNm5amM8LB84yPOol0REdf2GoMXpGqVIhMuD20NDWz6KidMTiibsu1a2LCxkY8bL5rVaro6hLpEVdqRA50tSLz4ydrRltVs52kZ2SwAsHtQsm2mhRVypEqht7yHLGU5gemWeQvhe7TbhlUT6v1LTS7/ZM/AQVMbSoKxUC3YMjHGvrZ2FhWtR1vZxz6+JC3B4fLx/RaQOiiRZ1pULgpcMtUdv1cs7S4kwK0xN54YB2wUQTLepKhcDG6iYykx0UZUTGQhiTYbMJNy4s4LX6dga0CyZqaFFXKsh6h0fZUd/OwsLoG/VyofXz8xjx+NhWqyciRQst6koF2Wu17Yx6DfMK0qyOMmXLSjLJSHawSfvVo8aEJx8ppS7Nq2+1kp7kYGZW5M7IeM4TuyZe6ajU5eTF6mauKD6J3Sbcs6I4DMnUZGlLXakg8vkMW462srYiB7sturtezplXkMbQqJeTHQNWR1EB0KKuVBAdONNNx8AI6+bmWh0laObkpRBnE2qaeq2OogKgRV2pINr8Vis2gbURvMLRpUqIs1OWk8KRpl6M0el4I50WdaWC6NWjrSwtziTTGVvrsM8rSKNrcJSWPrfVUdQEtKgrFSQtvcNUN/ZyXQx1vZxzWX4qALXNfRYnURPRoq5UkGw5OjajYSz1p5+TnuQgPy2R2hYt6pFOi7pSQfJaXTt5aQnM9bdqY01FXgonOwZ1gq8Ip0VdqSDw+QyvH+tgdXl21J9FejFz8lLxGsPr9e1WR1HvQYu6UkFwpKmXzoERri7PtjpKyMxyJRMfZ2OLThkQ0bSoKxUE2/2t11gu6nE2G2U5KWw92qZDGyOYFnWlgmBHfTsVeSnkpkXfghiXoiIvhcbuIY619VsdRV3EhEVdRGaKyGYRqRGRwyLyRf/9/yAijSKy33+5OfRxlYo8w6Nedh/v5Ory2Dnh6GIq8sYOAm85ql0wkSqQlroH+IoxZh6wErhfROb7H/u+MWaJ/7IxZCmVimB7Tnbh9vi4eo7L6ighl5kcT1mOk9fq9GBppJqwqBtjmowxe/3X+4AaoCjUwZSKFq/VtRNnE5aXxn5Rh7HjBruPd+L2eK2OosZxSX3qIlICXAHs8t/1BRE5KCIPi0hmkLMpFRV21LeztDiTlITpMZP16vJshka97DvVbXUUNY6Ai7qIpABPA18yxvQCPwLKgCVAE/Ddizxvg4hUiUhVW5v2w6nY0jUwQvXZHlbH8KiXC60sc2GTsV9mKvIEVNRFxMFYQX/cGPMMgDGmxRjjNcb4gJ8Ay8d7rjHmQWNMpTGmMicn9g8kqenl9WMdGANXz5k+RT0t0cHimRlvD+NUkSWQ0S8CPATUGGO+d979BedtdgdQHfx4SkW27fXtpCbEsXhGutVRwurq8mwOnumhd3jU6ijqAoG01FcDHwfWXTB88V9E5JCIHASuA74cyqBKRaLt9W2sLHMRZ59ep3ysLs/G6zPsaui0Ooq6wIRHdowx24HxJrPQIYxqWjvVMcjpziE+c/Vsq6OE3RXFGSQ6bOyob2f9/Dyr46jzTK/mhVJB9Fr92IH/6dSffk5CnJ3lpS5eq9PBD5FGi7pSk7Sjvp2C9ERmZzutjmKJ1WUujrUN0No7bHUUdZ7pMbBWqSDz+gw76jt43/y8mJ1q92Ke2HUKgJ6hsYOk39tUy+UzMiZ83j0rikOaS43RlrpSk3D4bA89Q6PTsuvlnIL0JBLibBxvH7A6ijqPFnWlJuHcGO1VZdO3qNttwixXMg1a1COKFnWlJmF7XTtz81PJSU2wOoqlSl1O2vrcusRdBNGirtQlGhrxUnWiK6YXxAhUaU4KgHbBRBA9UKrUBM4dGDynrqWPEa8Pj8+867HppigjiXj7WL/6oqLpdVZtpNKWulKXqL6tH7tNKHFNz6GM57PbhGJXMsfbdSWkSKFFXalLVN/aT3HW2CLMCmZnO2npdTOg/eoRQT+VSl2CfreHpp5hynNTrI4SMUr9J1+d6NB+9UigRV2pS9DgX3C5PEeL+jlFmUk47KJDGyOEFnWlLkF9az+JDhtFmUlWR4kYcTYbM7OSOaFFPSJoUVcqQMYY6lv7mZ2dgm2aTQ0wkdnZTpp7hhka0XVLraZFXakAdQyM0D00qv3p4yjNTsGg/eqRQIu6UgGqb/X3p2tRf5cZmUnE2eTtYw7KOlrUlQpQfWs/GUkOXM54q6NEHId9rF/9uLbULadFXakA+Iyhob2f8tyUaTfVbqBKs500dQ8zPKr96lbSoq5UABq7hhge9VGmXS8XVZrt1H71CKBFXakA1Pv7ist0fPpFFWclY7cJx9u0qFtpwqIuIjNFZLOI1IjIYRH5ov/+LBHZJCJ1/n8zQx9XKWvUt/ZTkJ5ISoLOgXcxDruNmZlJ2q9usUBa6h7gK8aYecBK4H4RmQ98FXjFGDMHeMV/W6mY4/Z4OdU5qGeRBqAk28nZ7iHcHu1Xt8qERd0Y02SM2eu/3gfUAEXAbcCj/s0eBW4PVUilrHSsdQCvz1CRn2p1lIhX4nLiM3Cqc9DqKNPWJfWpi0gJcAWwC8gzxjTBWOEHcoMdTqlIcLSll4Q4G7NcyVZHiXizspIR4ES7FnWrBFzURSQFeBr4kjGm9xKet0FEqkSkqq2tbTIZlbKMMYajzX2U56YQZ9NxBRNJcNgpzEjSETAWCuhTKiIOxgr648aYZ/x3t4hIgf/xAqB1vOcaYx40xlQaYypzcnKCkVmpsDnS1EvvsIe52vUSsBJXMqc7B/F4fVZHmZYCGf0iwENAjTHme+c99Dxwr//6vcBzwY+nlLU2vzXWVqnI06IeqJJsJx6fobF7yOoo01IgLfXVwMeBdSKy33+5GfgOsF5E6oD1/ttKxZTNR9soykgiNdFhdZSoMcu/zJ9OxWuNCQfdGmO2Axc7L/r64MZRKnJ0DYyw71QX116mYwAuRUpCHDmpCZzoGGSt1WGmIT3yo9RFbK1tw2fgMu16uWSlLicnOgbwGWN1lGlHi7pSF/G76mZyUxN0laNJKMlOxu3x0dwzbHWUaUeLulLjGBzxsKW2lRsX5usqR5NQ4tLFqK2iRV2pcWw92sbwqI8bF+ZbHSUqZSTHk5Hs0IOlFtCirtQ4XqxuJssZz/KSLKujRK1Sl5PjHYMY7VcPKy3qSl3A7fHy6lutvG9+HnF2/YpMVonLyYDbQ0f/iNVRphX9xCp1ge117fS7Pbxfu16mpCR7rF9dp+INLy3qSl3gxepmUhPjWF2WbXWUqJadEo8zIU771cNMi7pS5xke9fL7w82sn59HfJx+PaZCRChxJesImDDTT61S53m5poW+YQ8fumKG1VFiQonLSdfgKN2D2q8eLlrUlTrPM3sbKUhP5Koyl9VRYkJp9rnx6jq/erhoUVfKr63PzdbaNm6/ogi7TU84Cob89EQS4mzaBRNGWtSV8ntufyNen+HOpUVWR4kZNhFmuZL1YGkYaVFXyu+ZvY0snpFOea5O4BVMJS4nrX1uOge0Xz0ctKgrBRw+28ORpl7uuEJb6cF2bh6YN090WpxketCirhTw6OsnSHLYuUNHvQTdjMwk4mzCm8e1qIeDFnU17XX0u/n1/rN8aGkR6cm6wlGwxdltzMhM1pZ6mGhRV9Pek2+eZsTj45OrSqyOErNKs5OpPtvLgNtjdZSYp0VdTWujXh+P7TzJNXOymaMrHIVMicuJ12fYe6rL6igxT4u6mtZ+V91Mc++wttJDrDgrGbtN2K396iE3YVEXkYdFpFVEqs+77x9EpFFE9vsvN4c2plLB5/MZHthyjNJsJ9fp4tIhleCws6AwTYt6GATSUn8EuHGc+79vjFniv2wMbiylQu/F6mZqmnr54vVzsOkZpCG3vCSLfae7cXu8VkeJaRMWdWPMNkB/vaqY4vUZvv9yLXNyU7h1caHVcaaFZaVZjHh8HDrTY3WUmDaVPvUviMhBf/dM5sU2EpENIlIlIlVtbW1TeDmlguf5A43Ut/bz5fUVOs9LmCzzLw24S7tgQmqyRf1HQBmwBGgCvnuxDY0xDxpjKo0xlTk5OZN8OaWCx+3x8oOX65hfkMaNC3R1o3DJcsYzJzdFx6uH2KSKujGmxRjjNcb4gJ8Ay4MbS6nQ+fGWBk52DPLVm+ZqX3qYLS/NYs+JLrw+XYw6VCZV1EWk4LybdwDVF9tWqUjS0NbPDzfX88HFhayp0L8cw215aRZ9bg81Tb1WR4lZcRNtICL/A1wLZIvIGeDvgWtFZAlggBPAZ0OYUamgMMbwt89Wk+Cw8fUPzLM6zrR0rl999/FOFhalW5wmNk1Y1I0xHxnn7odCkEWpkHpi9yl2NnTw7TsWkpuaaHWcaakwI4kZmUm8eaKTT11danWcmKRnlKppobqxh2++cIRr5mTzkWXFVseZ1paXZrH7eCfGaL96KEzYUlcq2vUMjvK5x/fgcsazZk4OT7552upI09rykiye2dtIQ/sAZTkpVseJOdpSVzFt1Ovjz5/cR1P3MD/86FKcCdqOsdryUv949QYd2hgKWtRVzPL6DF9+aj9ba9v41m0LWVp80XPkVBiVZjvJS0tgx7F2q6PEJC3qKib5fIa/eeYQvznYxNdumss9K7QfPVKICKvLsnnjWAc+Ha8edPq3qLLUE7tOTep571Wkh0e9fOUXB/jtoSb+fF05n11bNtl4KkRWlWfzzL5Gjrb0Ma8gzeo4MUVb6iqmtPe7uecnb7Cxuom/vXkeX15fYXUkNY5VZS4AdtRrF0ywaVFXMeN31U3c+INtHD7bywP3LOVP1sxGRKcBiESFGUmUZjt5/ViH1VFijna/qKjXOTDCN56r5jcHm1hQmMZjn16sf9JHgVVlLn69r5FRrw+HXduXwaJ7UkW1Fw81sf57W/n94Wb+Yn0Fv75/tRb0KLG6PJuBES8HdX71oNKWuopKHf1uvvH8YX57sImFRWn8/MMrtJhHmZWzx/rVX69v58pZOtw0WLSoq6hzqLGH7750lN7hUf7yfRV8dm2Z/vkehbKc8cwvSGN7fTt/dv0cq+PEDC3qKmr0uz08f+As1Y09LCpK5/E/WsHcfG2dR7NrKrJ56LXj9Ls9pOjZvkGhzRsVFY429/GDl2upaerlffPzePbzq7Sgx4C1FTl4fIadOgomaPRXo4poXp9h05EWttW1kZ+WyF3LZpKflkicdrfEhMpZWSTH29la28r6+XlWx4kJWtRVxBoe9fLErlPUt/WzvCSLWy4v0L7zGBMfZ2NVmYuttW0YY/S8giDQb4iKSD1Dozy4rYGG9n7uXDqD268o0oIeo9ZW5HC6c4gTHYNWR4kJ+i1REadzYIQfbz1G1+AI964q0eFuMW5tRS4AW4+2WpwkNmhRVxGlZ2iUh7Y3MOLx8SfXzGZObqrVkVSIFbuSKXEls61O54EJBi3qKmL0DY8V9MERL/etLqEwI8nqSCpM1lbksPNYB26P1+ooUW/CA6Ui8jDwAaDVGLPQf18W8BRQApwA7jLGdIUupop1Hq+Px3edomdolE+tLmVGZvJ7bj/ZKXtVZLr2slwe3XmSNxo6WVuRY3WcqBZIS/0R4MYL7vsq8IoxZg7wiv+2UpNijOHX+xs51TnIh6+cySyX0+pIKsyuKnOR5LDz8pEWq6NEvQmLujFmG3DhYoK3AY/6rz8K3B7kXGoa2XGsg72nulk3N5dFRelWx1EWSHTYWVORzcs1LRijqyFNxWT71POMMU0A/n9zL7ahiGwQkSoRqWpra5vky6lYdbpzkN9VNzG/II11cy/6MVLTwA3z8mjqGebw2V6ro0S1kJ98ZIx5EHgQoLKyUn8Fq7f1DY/yVNVp0hId3Ll0BjY98SSmTXQcpN/tQYDvbarlhnljZ5fq2rKXbrIt9RYRKQDw/6sDTNUl+8Zzh+kaGOHuZTNJirdbHUdZLCUhjmJXMm81aUt9KiZb1J8H7vVfvxd4Ljhx1HTx4qEmnt3XyLq5uXpgVL1tXn4aZ3uG6R4csTpK1JqwqIvI/wA7gctE5IyIfBr4DrBeROqA9f7bSgWke3CEv3vuMAuL0rj2Mu1HV+84t9BJTXOfxUmi14R96saYj1zkoeuDnEVNE9/6zRG6B0f4708tZ//pbqvjqAiSk5pATmoC1Y09XOVfGUldGj2jVIXVlqOtPLO3kT9dW8b8Qp0PXb3boqJ0TrQP0Ds8anWUqKRFXYXN8KiXbzx3mNk5Tv7s+nKr46gItagoHQNUN+qC1JOh86mrsPnx1mOc6hzk8c+sICFOR7uo8eWlJZKXlsAhLeqToi11FRYnOwZ4YMsxPnB5AavLs62OoyLcoqIMTnYM0tQzZHWUqKNFXYXFN184gsMmfP2W+VZHUVHgcv90Eb892GRxkuij3S8qaC52xuDR5j5efauVmxbm8+pbep6amlh2agIF6Yn85mATn7lmttVxooq21FVIeX2GjdVNuJzxXFWmQ9RU4BbPyGD/6W4a2vqtjhJVtKirkNp1vIO2Pjc3LyogzqYfNxW4JTMzsAn8as8Zq6NEFf2WqZAZHPHwSk0rZTlO5ubrsnTq0qQlObj2slye3nsGr0/nAgyUFnUVMq+81crwqJebFxUgOgOjmoQ/unIGLb1uttXptN2B0qKuQqK1d5hdDR0sK8miIF3XGlWTc/28PDKTHfyqSrtgAqVFXYXEi9XNOOw2bpifZ3UUFcXi42zctqSITUdadObGAGlRV0FX29LH0ZY+1s3NJSVBR82qqbmrciYjXh9P7220OkpU0KKugsrrM/z2kA5hVMEzvzCNK2dl8tjOE/j0gOmEtKiroHqjQYcwquC7d1UJJzoG2VqrB0wnot86FTT9bg8v17QwJzdFhzCqoLppYT65qQn87PUTVkeJeFrUVdC8dLiZUa+PWy7XIYwquBx2Gx9dMYtttW0c0zNM35MWdRUUh870sOdkF1fNdpGbmmh1HBWD7llRjMMu/Le21t+TFnU1ZcYYvvnCYZLj7aybq0MYVWjkpCbwwcVFPFV1mo5+t9VxItaUirqInBCRQyKyX0SqghVKRZfnD5yl6mQX71uQT1K8Ln6hQudz15bh9vh4eMdxq6NErGC01K8zxiwxxlQG4WepKDPg9vBPG2tYWDQ27EypUCrPTeGmhfn89+sn6RnSNUzHo90vakr+c3M9Lb1u/uHWBdj04KgKg/uvK6fP7dG+9YuYalE3wEsiskdENgQjkIoeNU29/GRbA3cunUFlSZbVcdQ0saAwnXVzc3l4x3H63R6r40ScqRb11caYpcBNwP0isubCDURkg4hUiUhVW5ueOBArvD7D1545RFqSg6/fMs/qOGqa+fPr59A1OMpPX2uwOkrEmdLEHMaYs/5/W0XkWWA5sO2CbR4EHgSorKzUc3xjxM/fOMn+09384O4lZDrjrY6jYtTFlkgEWFCYxgNbjpHksJOa6PiDx+5ZURzqaBFr0i11EXGKSOq568D7gOpgBVOR61THIP/yu7dYU5HDbUsKrY6jpqn3z8/H4/Wx+aiue3u+qXS/5AHbReQAsBv4rTHmd8GJpSKV12f4i1/sxybC//vQIj1zVFkmOzWBypIsdh/vpF3Hrb9t0kXdGNNgjFnsvywwxnw7mMFUZHpwWwNVJ7v45m0LKMrQxS+Uta6fm0uczcaLh5qsjhIxdEijClh1Yw/f23SUmxflc8cVRVbHUYrURAfr5uZS09zHW029VseJCFrUVUB6Bkf5/ON7cTkT+L+3a7eLihyryl3kpibwwsGzjHh8VsexnBZ1NSGfz/CVX+6nqWeIBz62lCwd7aIiSJzNxgcXF9I1OMrWWj1oqkVdTeiHm+t5uaaVr98yn6XFOhWAijyzc1JYMjODbbXtNPUMWR3HUlrU1Xt6bn8j391Uyx1XFPGJq2ZZHUepi7plUQFJ8XZ+WXUGt8drdRzLaFFXF/V6fTt/+csDrJydxXfu1H50FdmcCXF86IoimnuH+beX66yOYxkt6mpc+09389nH9lCa7eS/Pl5JQpxOqasi39yCsdlCf7z1GG+e6LQ6jiW0qKt32XOyi4//dBeZzngeuW856UmOiZ+kVIS4ZVEBxVnJfOGJvdPypCQt6uoP7DzWwSce2kV2agJPfXYlhXqCkYoyiQ47D3z0SroHR/nik/vw+qbXlFNa1NXbflF1mk88vIuCjCSe3LCSgnQt6Co6zS9M4x9vX8iO+g7+/0tHrY4TVlOapVFFtvea4e58Xp/hpSPNvFbXTnlOCk9/bpV2uaiod1flTPad6uZHW45R4krm7mXTY+ZGLerTXNfgCE+9eZpTnYMsL83i1ssLtaCrmPGt2xbQ2D3E3zxbTX56EmsrcqyOFHLa/TJNGWPYd6qL/3i1jpbeYe6unMntS4qw23TYooodDruNH94XyXs9AAAIxUlEQVRzBRV5qXz+53vYc7LL6kghp0V9GuoaGOGR10/wyz1nyE1N5AvXlbN4ZobVsZQKidREB4/ct4yc1ATufXg3e07G9lBH7X6ZRtweL1tr29he147NJtx6eQErZrvetWB0oH3xSkWLvLREntxwFR/5yRt84qHd/PTeZVxV5rI6VkhoS30aGPX62FHfzvdeqmXL0TYWFqXzpevncFVZ9rsKulKxKj89kSc3rCQ/PZFPPLyLX1SdtjpSSGhLPYaNeHzsPt7Btrp2+t0eSrOdfHRFMcUup9XRlLJEXloiz3xuNfc/sZf/86uD1Lf281fvvwyHPXbat1rUY1B7v5un3jzNA5vrGRjxUpbj5CNziynN1mKuVHqyg5/dt4xvvXCEB7c18EZDB9+/ewllOSlWRwsKLeoxwhjDzoYOHt91ipcONzPqNZTnpnD93FxmactcqT/gsNv4x9sXsqrMxdeePcQt//4a919bzmeumU1SfHTPc6RFfQomc0DxnhXBPQHiePsAGw818fSeMzS0D5Ce5ODjK0u4Z8VMdh+P/eFbSk3FTYsKuHJWJn///GG+u6mWJ3af4ks3zOG2JUUkOqKzuE+pqIvIjcC/AXbgp8aY7wQlVYQb9frw+gw+Y8J+oHFwxMPek928VtfG1to23mruA6ByViZfWFfOzYsK3v4walFXamK5aYn86GNXsquhg29vrOGvnz7Ev/7+KPcsL+aDS4ooz42ubplJF3URsQM/BNYDZ4A3ReR5Y8yRYIULh+FRL+39btr6/Bf/9da+d+7rHBhhaNTL8IiXoVEvnvMmCLIJ2G2C3SYkxNlxxttJTogjOd6OMz6O5AT/v/F2kuPjqG7sIdMZT0aSg+R4+7hzlI96fXQPjtLSO0xzzzAN7f3Ut/ZzqLGX2pY+vD6Dwy5cOSuTv/vAfG5cmE+RTryl1JSsmO3iuftXs/NYBw9tP85/bK7n31+t57K8VK6dm8OK0iyuLM4iPTmyz7ieSkt9OVBvjGkAEJEngduAkBR1n79l7DUGn4/zrhs8PsPwqJchf9E99+/wqJcBt5euwRE6Bkbo7Pf/O+Cma3CUjn43vcOecV/P5YwnJzWBnNQEZrmSSY6PI9FhI8lhJ8lhx2YT9p7qwuszeP0Z3KM+Bkc8DLg9dA2MMDDiYXj0DxfCfXjH8T+4bROIs9uI8/9icHt84y6em50Sz7yCNG6YV8bS4kyWl2bhTNDeM6WCSURYVZ7NqvJsmnuGebG6iRerm3l4+3H+a2sDAPlpiVTkp1KQlkhuWgK5/jrhSkkYqw/x9rfrRHyc7e1GX5xNwrLQzFSqQhFw/kDPM8CKqcUZ39/9uprH3jg5pZ8RZxOynPFkOeNxpcRTlJmMyxlPdspY8c5NTXy7iGc54wMa4hRIn7rXZ8YK/cjYL5srZ2XSMzRC1+Aow6NePN6xXwhenw+Pz7zd2k9LcpCXlkheWgKl2U4yknWxZ6XCKT89kftWl3Lf6lKGRrzsO9XFgTM91LX0Udfaz1tNvbT3uwl0Zt+f3beM6y7LDW1oplbUx/uV8663JyIbgA3+m/0iYtk8mMesedlsoN2al444ui/G6H54R0j2xUeD/QODYN0/T7jJe+2LgBcInkpRPwPMPO/2DODshRsZYx4EHpzC60Q1EakyxlRanSMS6L4Yo/vhHbov3hGsfTGV06jeBOaISKmIxAN/DDw/1UBKKaUmb9ItdWOMR0S+APyesSGNDxtjDgctmVJKqUs2peETxpiNwMYgZYlV07braRy6L8bofniH7ot3BGVfiDHTa1FWpZSKZbEzNZlSSikt6pMlIjeKyFERqReRr47z+CwReUVEDorIFhGZcd5j94pInf9yb3iTB99k94WILBGRnSJy2P/Y3eFPH1xT+Vz4H08TkUYR+c/wpQ6NKX5HikXkJRGpEZEjIlISzuzBNMX98C/+70eNiPy7BHL2kjFGL5d4YezA8DFgNhAPHADmX7DNL4F7/dfXAY/5r2cBDf5/M/3XM61+Txbtiwpgjv96IdAEZFj9nqzYF+c9/m/AE8B/Wv1+rNwXwBZgvf96CpBs9XsK934AVgE7/D/DDuwErp3oNbWlPjlvT5FgjBkBzk2RcL75wCv+65vPe/z9wCZjTKcxpgvYBNwYhsyhMul9YYypNcbU+a+fBVqBaF7ufSqfC0TkSiAPeCkMWUNt0vtCROYDccaYTQDGmH5jzGB4YgfdVD4TBkhk7JdBAuAAWiZ6QS3qkzPeFAlFF2xzALjTf/0OIFVEXAE+N5pMZV+8TUSWM/bhtejE36CY9L4QERvwXeCvQp4yPKbyuagAukXkGRHZJyL/6p9AMBpNej8YY3YyVuSb/JffG2NqJnpBLeqTE8gUCX8JrBWRfcBaoBHwBPjcaDKVfTH2A0QKgMeA+4wx757NLHpMZV98HthojImVhTOnsi/igGv8jy9jrOvikyFLGlqT3g8iUg7MY+xs/SJgnYismegFdZq/yZlwigR/d8KHAEQkBbjTGNMjImeAay947pZQhg2xSe8L/+004LfA140xb4QlcehM5XNxFXCNiHyesT7keBHpN8a868BalJjqd2SfeWcG2F8DK4GHwhE8yKayHzYAbxhj+v2PvcjYftj2nq9o9YGEaLww9suwASjlnYMfCy7YJhuw+a9/G/iW/3oWcJyxg6SZ/utZVr8ni/ZFPGN9iV+y+n1YvS8u2OaTRP+B0ql8Luz+7XP8t38G3G/1e7JgP9wNvOz/GQ7/d+XWiV5Tu18mwRjjAc5NkVAD/MIYc1hEviUiH/Rvdi1wVERqGTv49W3/czuBf2Rs7pw3GfsP7AzzWwiaqewL4C5gDfBJEdnvvywJ7zsIninui5gyxe+Il7EuiVdE5BBjXRg/CfNbCIopfiZ+xdgxpkOM/TI4YIx5YaLX1DNKlVIqhmhLXSmlYogWdaWUiiFa1JVSKoZoUVdKqRiiRV0ppWKIFnWllIohWtSVUiqGaFFXSqkY8r9tYd8QDEw/TwAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Distribution of predicted correlations\n",
    "sns.distplot(correlations)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "mpg             1.000000\n",
       "cylinders      -0.821864\n",
       "displacement   -0.855692\n",
       "horsepower     -0.853616\n",
       "weight         -0.874947\n",
       "acceleration    0.438677\n",
       "model_year      0.573469\n",
       "Name: mpg, dtype: float64"
      ]
     },
     "execution_count": 31,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Sanity check: are predictions better than any single correlation?\n",
    "data.corr('spearman').loc['mpg', :]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<seaborn.matrix.ClusterGrid at 0x7f85b96b4c88>"
      ]
     },
     "execution_count": 32,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x720 with 4 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Feature importances\n",
    "sns.clustermap(feature_importances)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}