Random Forest Quartiles.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# NOTE: This method was adapted from Will Koehrsen's guide to implementing Random Forest in Python (https://towardsdatascience.com/random-forest-in-python-24d0893d51c0)\n",
    "\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "import statsmodels.api as sm\n",
    "import matplotlib.pyplot as plt\n",
    "import random\n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn.ensemble import RandomForestRegressor\n",
    "from sklearn.tree import export_graphviz\n",
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'Index,\\x82_\\x8fRank,Song,Artist,Year,Lyrics,ids,energy,liveness,tempo,speechiness,acousticness,instrumentalness,time_signature,danceability,key,duration_ms,loudness,valence,quartile\\r\\n'"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "f = open(\"data/FINAL_cleaned_output.csv\")\n",
    "f.readline()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "d = np.loadtxt(f, delimiter=\",\", usecols = (0, 1, 4, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "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>index</th>\n",
       "      <th>rank</th>\n",
       "      <th>year</th>\n",
       "      <th>energy</th>\n",
       "      <th>liveness</th>\n",
       "      <th>tempo</th>\n",
       "      <th>speechiness</th>\n",
       "      <th>acousticness</th>\n",
       "      <th>instrumentalness</th>\n",
       "      <th>timesig</th>\n",
       "      <th>danceability</th>\n",
       "      <th>key</th>\n",
       "      <th>duration</th>\n",
       "      <th>loudness</th>\n",
       "      <th>valence</th>\n",
       "      <th>quartiles</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>count</th>\n",
       "      <td>1274.000000</td>\n",
       "      <td>1274.000000</td>\n",
       "      <td>1274.000000</td>\n",
       "      <td>1274.000000</td>\n",
       "      <td>1274.000000</td>\n",
       "      <td>1274.000000</td>\n",
       "      <td>1274.000000</td>\n",
       "      <td>1274.000000</td>\n",
       "      <td>1274.000000</td>\n",
       "      <td>1274.000000</td>\n",
       "      <td>1274.000000</td>\n",
       "      <td>1274.000000</td>\n",
       "      <td>1274.000000</td>\n",
       "      <td>1274.000000</td>\n",
       "      <td>1274.000000</td>\n",
       "      <td>1274.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mean</th>\n",
       "      <td>785.591052</td>\n",
       "      <td>50.484301</td>\n",
       "      <td>2007.361068</td>\n",
       "      <td>0.690180</td>\n",
       "      <td>0.177648</td>\n",
       "      <td>120.143451</td>\n",
       "      <td>0.102686</td>\n",
       "      <td>0.157085</td>\n",
       "      <td>0.011261</td>\n",
       "      <td>3.982732</td>\n",
       "      <td>0.650279</td>\n",
       "      <td>5.140502</td>\n",
       "      <td>232058.603611</td>\n",
       "      <td>-5.906388</td>\n",
       "      <td>0.545966</td>\n",
       "      <td>2.502355</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>std</th>\n",
       "      <td>467.282818</td>\n",
       "      <td>29.357217</td>\n",
       "      <td>4.662524</td>\n",
       "      <td>0.165599</td>\n",
       "      <td>0.135908</td>\n",
       "      <td>28.494104</td>\n",
       "      <td>0.098734</td>\n",
       "      <td>0.199296</td>\n",
       "      <td>0.079076</td>\n",
       "      <td>0.253212</td>\n",
       "      <td>0.143762</td>\n",
       "      <td>3.553635</td>\n",
       "      <td>40722.787068</td>\n",
       "      <td>2.385913</td>\n",
       "      <td>0.224701</td>\n",
       "      <td>1.131736</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>min</th>\n",
       "      <td>0.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>2000.000000</td>\n",
       "      <td>0.016400</td>\n",
       "      <td>0.021000</td>\n",
       "      <td>57.107000</td>\n",
       "      <td>0.023600</td>\n",
       "      <td>0.000046</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.178000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>78200.000000</td>\n",
       "      <td>-32.580000</td>\n",
       "      <td>0.039600</td>\n",
       "      <td>1.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25%</th>\n",
       "      <td>375.250000</td>\n",
       "      <td>24.000000</td>\n",
       "      <td>2003.000000</td>\n",
       "      <td>0.584000</td>\n",
       "      <td>0.091500</td>\n",
       "      <td>96.971250</td>\n",
       "      <td>0.038000</td>\n",
       "      <td>0.017800</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>4.000000</td>\n",
       "      <td>0.556250</td>\n",
       "      <td>2.000000</td>\n",
       "      <td>206233.250000</td>\n",
       "      <td>-6.978000</td>\n",
       "      <td>0.371250</td>\n",
       "      <td>1.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>50%</th>\n",
       "      <td>784.500000</td>\n",
       "      <td>51.000000</td>\n",
       "      <td>2007.000000</td>\n",
       "      <td>0.709500</td>\n",
       "      <td>0.124000</td>\n",
       "      <td>119.969000</td>\n",
       "      <td>0.056850</td>\n",
       "      <td>0.069950</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>4.000000</td>\n",
       "      <td>0.662500</td>\n",
       "      <td>5.000000</td>\n",
       "      <td>227993.500000</td>\n",
       "      <td>-5.475000</td>\n",
       "      <td>0.557000</td>\n",
       "      <td>3.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>75%</th>\n",
       "      <td>1192.750000</td>\n",
       "      <td>76.000000</td>\n",
       "      <td>2011.000000</td>\n",
       "      <td>0.816750</td>\n",
       "      <td>0.225750</td>\n",
       "      <td>138.497250</td>\n",
       "      <td>0.131500</td>\n",
       "      <td>0.231000</td>\n",
       "      <td>0.000017</td>\n",
       "      <td>4.000000</td>\n",
       "      <td>0.749750</td>\n",
       "      <td>8.000000</td>\n",
       "      <td>252916.000000</td>\n",
       "      <td>-4.321250</td>\n",
       "      <td>0.729750</td>\n",
       "      <td>4.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>max</th>\n",
       "      <td>1598.000000</td>\n",
       "      <td>100.000000</td>\n",
       "      <td>2015.000000</td>\n",
       "      <td>0.987000</td>\n",
       "      <td>0.854000</td>\n",
       "      <td>210.750000</td>\n",
       "      <td>0.576000</td>\n",
       "      <td>0.996000</td>\n",
       "      <td>0.947000</td>\n",
       "      <td>5.000000</td>\n",
       "      <td>0.969000</td>\n",
       "      <td>11.000000</td>\n",
       "      <td>484147.000000</td>\n",
       "      <td>-1.097000</td>\n",
       "      <td>0.967000</td>\n",
       "      <td>4.000000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "             index         rank         year       energy     liveness  \\\n",
       "count  1274.000000  1274.000000  1274.000000  1274.000000  1274.000000   \n",
       "mean    785.591052    50.484301  2007.361068     0.690180     0.177648   \n",
       "std     467.282818    29.357217     4.662524     0.165599     0.135908   \n",
       "min       0.000000     1.000000  2000.000000     0.016400     0.021000   \n",
       "25%     375.250000    24.000000  2003.000000     0.584000     0.091500   \n",
       "50%     784.500000    51.000000  2007.000000     0.709500     0.124000   \n",
       "75%    1192.750000    76.000000  2011.000000     0.816750     0.225750   \n",
       "max    1598.000000   100.000000  2015.000000     0.987000     0.854000   \n",
       "\n",
       "             tempo  speechiness  acousticness  instrumentalness      timesig  \\\n",
       "count  1274.000000  1274.000000   1274.000000       1274.000000  1274.000000   \n",
       "mean    120.143451     0.102686      0.157085          0.011261     3.982732   \n",
       "std      28.494104     0.098734      0.199296          0.079076     0.253212   \n",
       "min      57.107000     0.023600      0.000046          0.000000     0.000000   \n",
       "25%      96.971250     0.038000      0.017800          0.000000     4.000000   \n",
       "50%     119.969000     0.056850      0.069950          0.000000     4.000000   \n",
       "75%     138.497250     0.131500      0.231000          0.000017     4.000000   \n",
       "max     210.750000     0.576000      0.996000          0.947000     5.000000   \n",
       "\n",
       "       danceability          key       duration     loudness      valence  \\\n",
       "count   1274.000000  1274.000000    1274.000000  1274.000000  1274.000000   \n",
       "mean       0.650279     5.140502  232058.603611    -5.906388     0.545966   \n",
       "std        0.143762     3.553635   40722.787068     2.385913     0.224701   \n",
       "min        0.178000     0.000000   78200.000000   -32.580000     0.039600   \n",
       "25%        0.556250     2.000000  206233.250000    -6.978000     0.371250   \n",
       "50%        0.662500     5.000000  227993.500000    -5.475000     0.557000   \n",
       "75%        0.749750     8.000000  252916.000000    -4.321250     0.729750   \n",
       "max        0.969000    11.000000  484147.000000    -1.097000     0.967000   \n",
       "\n",
       "         quartiles  \n",
       "count  1274.000000  \n",
       "mean      2.502355  \n",
       "std       1.131736  \n",
       "min       1.000000  \n",
       "25%       1.000000  \n",
       "50%       3.000000  \n",
       "75%       4.000000  \n",
       "max       4.000000  "
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "features = pd.DataFrame(data=d)\n",
    "features.columns = ['index','rank',\"year\",\"energy\",\"liveness\",\"tempo\",\"speechiness\",\"acousticness\",\"instrumentalness\",\"timesig\",\"danceability\",\"key\",\"duration\",\"loudness\",\"valence\",\"quartiles\"]\n",
    "features.describe()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "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>energy</th>\n",
       "      <th>liveness</th>\n",
       "      <th>tempo</th>\n",
       "      <th>speechiness</th>\n",
       "      <th>acousticness</th>\n",
       "      <th>instrumentalness</th>\n",
       "      <th>timesig</th>\n",
       "      <th>danceability</th>\n",
       "      <th>key</th>\n",
       "      <th>duration</th>\n",
       "      <th>loudness</th>\n",
       "      <th>valence</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>count</th>\n",
       "      <td>1274.000000</td>\n",
       "      <td>1274.000000</td>\n",
       "      <td>1274.000000</td>\n",
       "      <td>1274.000000</td>\n",
       "      <td>1274.000000</td>\n",
       "      <td>1274.000000</td>\n",
       "      <td>1274.000000</td>\n",
       "      <td>1274.000000</td>\n",
       "      <td>1274.000000</td>\n",
       "      <td>1274.000000</td>\n",
       "      <td>1274.000000</td>\n",
       "      <td>1274.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mean</th>\n",
       "      <td>0.690180</td>\n",
       "      <td>0.177648</td>\n",
       "      <td>120.143451</td>\n",
       "      <td>0.102686</td>\n",
       "      <td>0.157085</td>\n",
       "      <td>0.011261</td>\n",
       "      <td>3.982732</td>\n",
       "      <td>0.650279</td>\n",
       "      <td>5.140502</td>\n",
       "      <td>232058.603611</td>\n",
       "      <td>-5.906388</td>\n",
       "      <td>0.545966</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>std</th>\n",
       "      <td>0.165599</td>\n",
       "      <td>0.135908</td>\n",
       "      <td>28.494104</td>\n",
       "      <td>0.098734</td>\n",
       "      <td>0.199296</td>\n",
       "      <td>0.079076</td>\n",
       "      <td>0.253212</td>\n",
       "      <td>0.143762</td>\n",
       "      <td>3.553635</td>\n",
       "      <td>40722.787068</td>\n",
       "      <td>2.385913</td>\n",
       "      <td>0.224701</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>min</th>\n",
       "      <td>0.016400</td>\n",
       "      <td>0.021000</td>\n",
       "      <td>57.107000</td>\n",
       "      <td>0.023600</td>\n",
       "      <td>0.000046</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.178000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>78200.000000</td>\n",
       "      <td>-32.580000</td>\n",
       "      <td>0.039600</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25%</th>\n",
       "      <td>0.584000</td>\n",
       "      <td>0.091500</td>\n",
       "      <td>96.971250</td>\n",
       "      <td>0.038000</td>\n",
       "      <td>0.017800</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>4.000000</td>\n",
       "      <td>0.556250</td>\n",
       "      <td>2.000000</td>\n",
       "      <td>206233.250000</td>\n",
       "      <td>-6.978000</td>\n",
       "      <td>0.371250</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>50%</th>\n",
       "      <td>0.709500</td>\n",
       "      <td>0.124000</td>\n",
       "      <td>119.969000</td>\n",
       "      <td>0.056850</td>\n",
       "      <td>0.069950</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>4.000000</td>\n",
       "      <td>0.662500</td>\n",
       "      <td>5.000000</td>\n",
       "      <td>227993.500000</td>\n",
       "      <td>-5.475000</td>\n",
       "      <td>0.557000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>75%</th>\n",
       "      <td>0.816750</td>\n",
       "      <td>0.225750</td>\n",
       "      <td>138.497250</td>\n",
       "      <td>0.131500</td>\n",
       "      <td>0.231000</td>\n",
       "      <td>0.000017</td>\n",
       "      <td>4.000000</td>\n",
       "      <td>0.749750</td>\n",
       "      <td>8.000000</td>\n",
       "      <td>252916.000000</td>\n",
       "      <td>-4.321250</td>\n",
       "      <td>0.729750</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>max</th>\n",
       "      <td>0.987000</td>\n",
       "      <td>0.854000</td>\n",
       "      <td>210.750000</td>\n",
       "      <td>0.576000</td>\n",
       "      <td>0.996000</td>\n",
       "      <td>0.947000</td>\n",
       "      <td>5.000000</td>\n",
       "      <td>0.969000</td>\n",
       "      <td>11.000000</td>\n",
       "      <td>484147.000000</td>\n",
       "      <td>-1.097000</td>\n",
       "      <td>0.967000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "            energy     liveness        tempo  speechiness  acousticness  \\\n",
       "count  1274.000000  1274.000000  1274.000000  1274.000000   1274.000000   \n",
       "mean      0.690180     0.177648   120.143451     0.102686      0.157085   \n",
       "std       0.165599     0.135908    28.494104     0.098734      0.199296   \n",
       "min       0.016400     0.021000    57.107000     0.023600      0.000046   \n",
       "25%       0.584000     0.091500    96.971250     0.038000      0.017800   \n",
       "50%       0.709500     0.124000   119.969000     0.056850      0.069950   \n",
       "75%       0.816750     0.225750   138.497250     0.131500      0.231000   \n",
       "max       0.987000     0.854000   210.750000     0.576000      0.996000   \n",
       "\n",
       "       instrumentalness      timesig  danceability          key  \\\n",
       "count       1274.000000  1274.000000   1274.000000  1274.000000   \n",
       "mean           0.011261     3.982732      0.650279     5.140502   \n",
       "std            0.079076     0.253212      0.143762     3.553635   \n",
       "min            0.000000     0.000000      0.178000     0.000000   \n",
       "25%            0.000000     4.000000      0.556250     2.000000   \n",
       "50%            0.000000     4.000000      0.662500     5.000000   \n",
       "75%            0.000017     4.000000      0.749750     8.000000   \n",
       "max            0.947000     5.000000      0.969000    11.000000   \n",
       "\n",
       "            duration     loudness      valence  \n",
       "count    1274.000000  1274.000000  1274.000000  \n",
       "mean   232058.603611    -5.906388     0.545966  \n",
       "std     40722.787068     2.385913     0.224701  \n",
       "min     78200.000000   -32.580000     0.039600  \n",
       "25%    206233.250000    -6.978000     0.371250  \n",
       "50%    227993.500000    -5.475000     0.557000  \n",
       "75%    252916.000000    -4.321250     0.729750  \n",
       "max    484147.000000    -1.097000     0.967000  "
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "labels = np.array(features['quartiles'])\n",
    "original_features = list(features.columns)\n",
    "features = features.drop('rank', axis=1)\n",
    "features = features.drop('year', axis=1)\n",
    "features = features.drop('index', axis=1)\n",
    "features = features.drop('quartiles', axis=1)\n",
    "features.describe()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "feature_list = list(features.columns)\n",
    "\n",
    "features = np.array(features)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "train_features, test_features, train_labels, test_labels = train_test_split(features, labels, test_size = 0.2, random_state = 42)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "('Training Features Shape:', (1019, 12))\n",
      "('Training Labels Shape:', (1019,))\n",
      "('Testing Features Shape:', (255, 12))\n",
      "('Testing Labels Shape:', (255,))\n"
     ]
    }
   ],
   "source": [
    "print('Training Features Shape:', train_features.shape)\n",
    "print('Training Labels Shape:', train_labels.shape)\n",
    "print('Testing Features Shape:', test_features.shape)\n",
    "print('Testing Labels Shape:', test_labels.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "baseline_preds = [random.randint(1,4)*1.0 for i in range(len(test_features))]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "('Average baseline error: ', 1.25)\n"
     ]
    }
   ],
   "source": [
    "baseline_errors = abs(baseline_preds - test_labels)\n",
    "print('Average baseline error: ', round(np.mean(baseline_errors), 2))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "RandomForestRegressor(bootstrap=True, criterion='mse', max_depth=None,\n",
       "           max_features='auto', max_leaf_nodes=None,\n",
       "           min_impurity_decrease=0.0, min_impurity_split=None,\n",
       "           min_samples_leaf=1, min_samples_split=2,\n",
       "           min_weight_fraction_leaf=0.0, n_estimators=1000, n_jobs=1,\n",
       "           oob_score=False, random_state=42, verbose=0, warm_start=False)"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "rf = RandomForestRegressor(n_estimators = 1000, random_state = 42)\n",
    "rf.fit(train_features, train_labels)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "('Mean Absolute Error:', 0.98)\n"
     ]
    }
   ],
   "source": [
    "predictions = rf.predict(test_features)\n",
    "errors = abs(predictions - test_labels)\n",
    "print('Mean Absolute Error:', round(np.mean(errors), 2))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "('Accuracy:', 46.16, '%.')\n"
     ]
    }
   ],
   "source": [
    "mean_absolute_percent_error = 100 * (errors / test_labels)\n",
    "accuracy = 100 - np.mean(mean_absolute_percent_error)\n",
    "print('Accuracy:', round(accuracy, 2), '%.')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Variable: danceability         Importance: 0.12\n",
      "Variable: speechiness          Importance: 0.11\n",
      "Variable: liveness             Importance: 0.1\n",
      "Variable: tempo                Importance: 0.1\n",
      "Variable: acousticness         Importance: 0.1\n",
      "Variable: duration             Importance: 0.1\n",
      "Variable: loudness             Importance: 0.1\n",
      "Variable: valence              Importance: 0.1\n",
      "Variable: energy               Importance: 0.09\n",
      "Variable: key                  Importance: 0.06\n",
      "Variable: instrumentalness     Importance: 0.04\n",
      "Variable: timesig              Importance: 0.0\n"
     ]
    }
   ],
   "source": [
    "importances = list(rf.feature_importances_)\n",
    "feature_importances = [(feature, round(importance, 2)) for feature, importance in zip(feature_list, importances)]\n",
    "feature_importances = sorted(feature_importances, key = lambda x: x[1], reverse = True)\n",
    "\n",
    "sorted_features = []\n",
    "sorted_importances = []\n",
    "for pair in feature_importances:\n",
    "    print('Variable: {:20} Importance: {}'.format(*pair))\n",
    "    sorted_features.append(pair[0])\n",
    "    sorted_importances.append(pair[1])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x1c0fff0a50>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.style.use('fivethirtyeight')\n",
    "x_values = list(range(len(importances)))\n",
    "plt.barh(x_values, list(reversed(sorted_importances)))\n",
    "plt.yticks(x_values, list(reversed(sorted_features)), rotation='horizontal')\n",
    "plt.xlabel('Importance'); plt.ylabel('Variable'); plt.title('Variable Importances');"
   ]
  }
 ],
 "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": "3.6.2"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}