05a_modeling-with-extracted-features.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "from sklearn.model_selection import train_test_split, cross_val_score, GridSearchCV\n",
    "from sklearn.svm import SVC\n",
    "from sklearn.ensemble import AdaBoostClassifier, BaggingClassifier, VotingClassifier, RandomForestClassifier\n",
    "from sklearn.pipeline import Pipeline\n",
    "from sklearn.tree import ExtraTreeClassifier, DecisionTreeClassifier\n",
    "from sklearn.preprocessing import StandardScaler, OneHotEncoder, PolynomialFeatures\n",
    "from sklearn.metrics import confusion_matrix, plot_confusion_matrix\n",
    "from tensorflow.keras.models import Sequential\n",
    "from tensorflow.keras.layers import Dense, Dropout\n",
    "from tensorflow.keras.utils import to_categorical\n",
    "from ast import literal_eval\n",
    "from tqdm import tqdm"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\ProgramData\\Anaconda3\\envs\\dsi\\lib\\site-packages\\tqdm\\std.py:670: FutureWarning: The Panel class is removed from pandas. Accessing it from the top-level namespace will also be removed in the next version\n",
      "  from pandas import Panel\n"
     ]
    }
   ],
   "source": [
    "tqdm.pandas()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "songs = pd.read_csv('./data/five_g_500_modeling.csv')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "features = songs.columns[22:].drop([])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "X = songs[features]\n",
    "y = songs['genre']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "def running_models(X, y, scaled=False, c=10):\n",
    "    X_train, X_test, y_train, y_test = train_test_split(X, y, stratify=y, random_state=182)\n",
    "    \n",
    "    # MODELING WITH SCALED FEATURES #\n",
    "    \n",
    "    # Scale the features\n",
    "    sc = StandardScaler()\n",
    "    X_train_sc = sc.fit_transform(X_train)\n",
    "    X_test_sc = sc.transform(X_test)\n",
    "    \n",
    "    # Creating a df to display results\n",
    "    model_df = pd.DataFrame(columns=['train_score', 'test_score'])\n",
    "    \n",
    "    model_sc = SVC(C=c, kernel='rbf', degree=3)\n",
    "    model_sc.fit(X_train_sc, y_train)\n",
    "    train_score = model_sc.score(X_train_sc, y_train)\n",
    "    test_score = model_sc.score(X_test_sc, y_test)\n",
    "    \n",
    "    row = {'model': type(model_sc).__name__,\n",
    "           'train_score' : train_score,\n",
    "          'test_score' : test_score,\n",
    "           'C': c,\n",
    "          'scaled': True}\n",
    "    model_df = model_df.append(row, ignore_index=True)\n",
    "    \n",
    "    # MODELING WITH UNSCALED FEATURES #\n",
    "    \n",
    "    model = SVC(C=c, kernel='rbf', degree=3)\n",
    "    model.fit(X_train, y_train)\n",
    "    train_score = model.score(X_train, y_train)\n",
    "    test_score = model.score(X_test, y_test)\n",
    "    \n",
    "    row = {'model': type(model).__name__,\n",
    "       'train_score' : train_score,\n",
    "      'test_score' : test_score,\n",
    "       'C': c,\n",
    "      'scaled': False}\n",
    "    model_df = model_df.append(row, ignore_index=True)\n",
    "    return model_df"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "There's a high **positive correlation between loudness and energy** and a high **negative correlation between energy and acousticness**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "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>train_score</th>\n",
       "      <th>test_score</th>\n",
       "      <th>C</th>\n",
       "      <th>model</th>\n",
       "      <th>scaled</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0.957867</td>\n",
       "      <td>0.8272</td>\n",
       "      <td>10.0</td>\n",
       "      <td>SVC</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>0.820267</td>\n",
       "      <td>0.8416</td>\n",
       "      <td>10.0</td>\n",
       "      <td>SVC</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   train_score  test_score     C model  scaled\n",
       "0     0.957867      0.8272  10.0   SVC     1.0\n",
       "1     0.820267      0.8416  10.0   SVC     0.0"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "running_models(X, y)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 523,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Create train_test_split\n",
    "X_train, X_test, y_train, y_test = train_test_split(X, y, stratify=y, random_state=182)\n",
    "\n",
    "# Scale the features\n",
    "sc = StandardScaler()\n",
    "X_train_sc = sc.fit_transform(X_train)\n",
    "X_test_sc = sc.transform(X_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 524,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "SVC(C=0.35)"
      ]
     },
     "execution_count": 524,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model = SVC(C=.35)\n",
    "model.fit(X_train_sc, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 525,
   "metadata": {},
   "outputs": [],
   "source": [
    "categories = list(set(y_preds))\n",
    "numbers = list(range(5))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 526,
   "metadata": {},
   "outputs": [],
   "source": [
    "y_preds = model.predict(X_test_sc)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 527,
   "metadata": {},
   "outputs": [],
   "source": [
    "results = X_test.copy()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 529,
   "metadata": {},
   "outputs": [],
   "source": [
    "results.loc[:, 'genre'] = y_test.copy()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 530,
   "metadata": {},
   "outputs": [],
   "source": [
    "results.loc[:, 'preds'] = y_preds.copy()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 531,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_confusion_matrix(model, X_test_sc, y_test, xticks_rotation='vertical', normalize='true');"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We can see here that our model performs **best on classical music**, and performs **worst on r&b, only predicting correctly 74% of the time.** I want to dig more into why that is"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 532,
   "metadata": {},
   "outputs": [],
   "source": [
    "results_wrong = results[results['genre'] != results['preds']]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 533,
   "metadata": {},
   "outputs": [],
   "source": [
    "songs_wrong = songs.loc[results_wrong.index, ['track_name']]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 534,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\ProgramData\\Anaconda3\\envs\\dsi\\lib\\site-packages\\ipykernel_launcher.py:1: SettingWithCopyWarning: \n",
      "A value is trying to be set on a copy of a slice from a DataFrame.\n",
      "Try using .loc[row_indexer,col_indexer] = value instead\n",
      "\n",
      "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
      "  \"\"\"Entry point for launching an IPython kernel.\n"
     ]
    }
   ],
   "source": [
    "results_wrong['track_name'] = songs_wrong"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 535,
   "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>mean_energy</th>\n",
       "      <th>std_energy</th>\n",
       "      <th>min_energy</th>\n",
       "      <th>max_energy</th>\n",
       "      <th>max_rms</th>\n",
       "      <th>std_rms</th>\n",
       "      <th>median_rms</th>\n",
       "      <th>min_rms</th>\n",
       "      <th>mfcc_mean_00</th>\n",
       "      <th>mfcc_mean_01</th>\n",
       "      <th>...</th>\n",
       "      <th>mfcc_min_05</th>\n",
       "      <th>mfcc_min_06</th>\n",
       "      <th>mfcc_min_07</th>\n",
       "      <th>mfcc_min_08</th>\n",
       "      <th>mfcc_min_09</th>\n",
       "      <th>mfcc_min_10</th>\n",
       "      <th>mfcc_min_11</th>\n",
       "      <th>genre</th>\n",
       "      <th>preds</th>\n",
       "      <th>track_name</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>1653</th>\n",
       "      <td>26.837427</td>\n",
       "      <td>29.822205</td>\n",
       "      <td>0.227380</td>\n",
       "      <td>231.272377</td>\n",
       "      <td>0.672089</td>\n",
       "      <td>0.110432</td>\n",
       "      <td>0.177969</td>\n",
       "      <td>0.021074</td>\n",
       "      <td>-132.137619</td>\n",
       "      <td>106.776787</td>\n",
       "      <td>...</td>\n",
       "      <td>-12.309896</td>\n",
       "      <td>-47.899212</td>\n",
       "      <td>-35.528416</td>\n",
       "      <td>-38.607246</td>\n",
       "      <td>-34.404018</td>\n",
       "      <td>-37.107895</td>\n",
       "      <td>-35.714878</td>\n",
       "      <td>rap</td>\n",
       "      <td>r&amp;b</td>\n",
       "      <td>Only You (feat. The Notorious B.I.G. &amp; Mase) -...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1778</th>\n",
       "      <td>46.566456</td>\n",
       "      <td>41.379618</td>\n",
       "      <td>0.605496</td>\n",
       "      <td>280.619677</td>\n",
       "      <td>0.740328</td>\n",
       "      <td>0.110651</td>\n",
       "      <td>0.267670</td>\n",
       "      <td>0.034389</td>\n",
       "      <td>-109.434654</td>\n",
       "      <td>109.749611</td>\n",
       "      <td>...</td>\n",
       "      <td>-4.021262</td>\n",
       "      <td>-65.765266</td>\n",
       "      <td>-25.861614</td>\n",
       "      <td>-27.706072</td>\n",
       "      <td>-78.043472</td>\n",
       "      <td>-24.386654</td>\n",
       "      <td>-23.708405</td>\n",
       "      <td>rap</td>\n",
       "      <td>r&amp;b</td>\n",
       "      <td>Somethin Tells Me</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1856</th>\n",
       "      <td>39.402694</td>\n",
       "      <td>30.653769</td>\n",
       "      <td>2.283467</td>\n",
       "      <td>236.971450</td>\n",
       "      <td>0.680320</td>\n",
       "      <td>0.092538</td>\n",
       "      <td>0.251618</td>\n",
       "      <td>0.066782</td>\n",
       "      <td>-85.101601</td>\n",
       "      <td>125.324379</td>\n",
       "      <td>...</td>\n",
       "      <td>-11.752495</td>\n",
       "      <td>-47.426876</td>\n",
       "      <td>-22.385229</td>\n",
       "      <td>-40.177628</td>\n",
       "      <td>-33.128696</td>\n",
       "      <td>-28.820669</td>\n",
       "      <td>-27.308964</td>\n",
       "      <td>rap</td>\n",
       "      <td>r&amp;b</td>\n",
       "      <td>Sky's the Limit (feat. 112) - 2014 Remaster</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1673</th>\n",
       "      <td>42.095754</td>\n",
       "      <td>27.215504</td>\n",
       "      <td>0.074239</td>\n",
       "      <td>172.595618</td>\n",
       "      <td>0.580604</td>\n",
       "      <td>0.101780</td>\n",
       "      <td>0.283009</td>\n",
       "      <td>0.012042</td>\n",
       "      <td>-77.639091</td>\n",
       "      <td>128.701294</td>\n",
       "      <td>...</td>\n",
       "      <td>-22.018921</td>\n",
       "      <td>-56.880360</td>\n",
       "      <td>-39.439629</td>\n",
       "      <td>-40.098961</td>\n",
       "      <td>-32.648499</td>\n",
       "      <td>-37.144257</td>\n",
       "      <td>-33.395630</td>\n",
       "      <td>rap</td>\n",
       "      <td>r&amp;b</td>\n",
       "      <td>Whatever You Like</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1538</th>\n",
       "      <td>45.313658</td>\n",
       "      <td>61.220352</td>\n",
       "      <td>0.203308</td>\n",
       "      <td>384.215485</td>\n",
       "      <td>0.866268</td>\n",
       "      <td>0.150881</td>\n",
       "      <td>0.250395</td>\n",
       "      <td>0.019927</td>\n",
       "      <td>-146.169693</td>\n",
       "      <td>117.854248</td>\n",
       "      <td>...</td>\n",
       "      <td>-23.983143</td>\n",
       "      <td>-59.268829</td>\n",
       "      <td>-26.802391</td>\n",
       "      <td>-56.876747</td>\n",
       "      <td>-33.827942</td>\n",
       "      <td>-16.999588</td>\n",
       "      <td>-41.450058</td>\n",
       "      <td>rap</td>\n",
       "      <td>r&amp;b</td>\n",
       "      <td>Can't Believe It (feat. Lil' Wayne)</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 59 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "      mean_energy  std_energy  min_energy  max_energy   max_rms   std_rms  \\\n",
       "1653    26.837427   29.822205    0.227380  231.272377  0.672089  0.110432   \n",
       "1778    46.566456   41.379618    0.605496  280.619677  0.740328  0.110651   \n",
       "1856    39.402694   30.653769    2.283467  236.971450  0.680320  0.092538   \n",
       "1673    42.095754   27.215504    0.074239  172.595618  0.580604  0.101780   \n",
       "1538    45.313658   61.220352    0.203308  384.215485  0.866268  0.150881   \n",
       "\n",
       "      median_rms   min_rms  mfcc_mean_00  mfcc_mean_01  ...  mfcc_min_05  \\\n",
       "1653    0.177969  0.021074   -132.137619    106.776787  ...   -12.309896   \n",
       "1778    0.267670  0.034389   -109.434654    109.749611  ...    -4.021262   \n",
       "1856    0.251618  0.066782    -85.101601    125.324379  ...   -11.752495   \n",
       "1673    0.283009  0.012042    -77.639091    128.701294  ...   -22.018921   \n",
       "1538    0.250395  0.019927   -146.169693    117.854248  ...   -23.983143   \n",
       "\n",
       "      mfcc_min_06  mfcc_min_07  mfcc_min_08  mfcc_min_09  mfcc_min_10  \\\n",
       "1653   -47.899212   -35.528416   -38.607246   -34.404018   -37.107895   \n",
       "1778   -65.765266   -25.861614   -27.706072   -78.043472   -24.386654   \n",
       "1856   -47.426876   -22.385229   -40.177628   -33.128696   -28.820669   \n",
       "1673   -56.880360   -39.439629   -40.098961   -32.648499   -37.144257   \n",
       "1538   -59.268829   -26.802391   -56.876747   -33.827942   -16.999588   \n",
       "\n",
       "      mfcc_min_11  genre  preds  \\\n",
       "1653   -35.714878    rap    r&b   \n",
       "1778   -23.708405    rap    r&b   \n",
       "1856   -27.308964    rap    r&b   \n",
       "1673   -33.395630    rap    r&b   \n",
       "1538   -41.450058    rap    r&b   \n",
       "\n",
       "                                             track_name  \n",
       "1653  Only You (feat. The Notorious B.I.G. & Mase) -...  \n",
       "1778                                  Somethin Tells Me  \n",
       "1856        Sky's the Limit (feat. 112) - 2014 Remaster  \n",
       "1673                                  Whatever You Like  \n",
       "1538                Can't Believe It (feat. Lil' Wayne)  \n",
       "\n",
       "[5 rows x 59 columns]"
      ]
     },
     "execution_count": 535,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "results_wrong[(results_wrong['preds'] == 'r&b') & (results_wrong['genre'] == 'rap')].head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 536,
   "metadata": {},
   "outputs": [],
   "source": [
    "group_preds = results_wrong.groupby('preds').mean()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 537,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Index(['min_rms', 'mfcc_mean_01', 'mfcc_mean_02', 'mfcc_mean_03',\n",
       "       'mfcc_mean_04', 'mfcc_mean_06', 'mfcc_mean_07', 'mfcc_mean_08',\n",
       "       'mfcc_mean_09', 'mfcc_min_06', 'mfcc_min_08'],\n",
       "      dtype='object')"
      ]
     },
     "execution_count": 537,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Getting columns where rap and r&b are very close to each other\n",
    "group_preds.columns[abs(group_preds.loc['rap']) - abs(group_preds.loc['r&b']) < 0.001]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 516,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "r&b                      0.333333\n",
       "rap                      0.254902\n",
       "progressive bluegrass    0.186275\n",
       "rock                     0.166667\n",
       "classical                0.058824\n",
       "Name: genre, dtype: float64"
      ]
     },
     "execution_count": 516,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "results_wrong['genre'].value_counts(normalize=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Rap and R&B seem to be predicted wrong most often with these new features I've pulled"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 517,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "r&b    1.0\n",
       "Name: preds, dtype: float64"
      ]
     },
     "execution_count": 517,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "results_wrong[results_wrong['genre'] == 'rap']['preds'].value_counts(normalize=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "It seems that when Rap is predicted wrong, it's predicted as r&b."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 518,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "rap                      0.852941\n",
       "rock                     0.088235\n",
       "progressive bluegrass    0.058824\n",
       "Name: preds, dtype: float64"
      ]
     },
     "execution_count": 518,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "results_wrong[results_wrong['genre'] == 'r&b']['preds'].value_counts(normalize=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "When R&B is predicted wrong, it's predicted to be rap 84% of the time"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 237,
   "metadata": {},
   "outputs": [],
   "source": [
    "genre_mean = songs.groupby('genre').mean()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Interesting that most of the genres hover slightly above 120 as a tempo. There's some interesting reasoning behind this with commercial music and the need for 30-seconds of music to fill an advertising slot. 128 beats per minute would allow for a 16-bar phrase in 4/4 time."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 241,
   "metadata": {},
   "outputs": [],
   "source": [
    "X = songs[features]\n",
    "# Thanks joris on SO: https://stackoverflow.com/questions/32011359/convert-categorical-data-in-pandas-dataframe/32011969\n",
    "y_cat = songs['genre'].astype('category').cat.codes"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 242,
   "metadata": {},
   "outputs": [],
   "source": [
    "y = to_categorical(y_cat)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 243,
   "metadata": {},
   "outputs": [],
   "source": [
    "X_train, X_test, y_train, y_test = train_test_split(X, y, stratify=y, random_state=182)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 244,
   "metadata": {},
   "outputs": [],
   "source": [
    "sc = StandardScaler()\n",
    "X_train_sc = sc.fit_transform(X_train)\n",
    "X_test_sc = sc.transform(X_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 299,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/100\n",
      "15/15 [==============================] - 0s 7ms/step - loss: 1.5583 - acc: 0.2448 - val_loss: 1.4482 - val_acc: 0.4304\n",
      "Epoch 2/100\n",
      "15/15 [==============================] - 0s 3ms/step - loss: 1.3683 - acc: 0.4475 - val_loss: 1.2088 - val_acc: 0.5952\n",
      "Epoch 3/100\n",
      "15/15 [==============================] - 0s 2ms/step - loss: 1.1525 - acc: 0.5424 - val_loss: 0.9566 - val_acc: 0.6800\n",
      "Epoch 4/100\n",
      "15/15 [==============================] - 0s 3ms/step - loss: 0.9417 - acc: 0.6347 - val_loss: 0.7834 - val_acc: 0.7024\n",
      "Epoch 5/100\n",
      "15/15 [==============================] - 0s 3ms/step - loss: 0.8217 - acc: 0.6560 - val_loss: 0.6928 - val_acc: 0.6976\n",
      "Epoch 6/100\n",
      "15/15 [==============================] - 0s 3ms/step - loss: 0.7537 - acc: 0.6843 - val_loss: 0.6420 - val_acc: 0.7264\n",
      "Epoch 7/100\n",
      "15/15 [==============================] - 0s 3ms/step - loss: 0.6976 - acc: 0.7029 - val_loss: 0.6025 - val_acc: 0.7344\n",
      "Epoch 8/100\n",
      "15/15 [==============================] - 0s 3ms/step - loss: 0.6543 - acc: 0.7200 - val_loss: 0.5836 - val_acc: 0.7568\n",
      "Epoch 9/100\n",
      "15/15 [==============================] - 0s 2ms/step - loss: 0.6317 - acc: 0.7291 - val_loss: 0.5556 - val_acc: 0.7680\n",
      "Epoch 10/100\n",
      "15/15 [==============================] - 0s 3ms/step - loss: 0.5939 - acc: 0.7541 - val_loss: 0.5349 - val_acc: 0.7808\n",
      "Epoch 11/100\n",
      "15/15 [==============================] - 0s 3ms/step - loss: 0.5797 - acc: 0.7611 - val_loss: 0.5220 - val_acc: 0.7888\n",
      "Epoch 12/100\n",
      "15/15 [==============================] - 0s 3ms/step - loss: 0.5547 - acc: 0.7712 - val_loss: 0.5128 - val_acc: 0.7920\n",
      "Epoch 13/100\n",
      "15/15 [==============================] - 0s 3ms/step - loss: 0.5437 - acc: 0.7781 - val_loss: 0.5069 - val_acc: 0.7904\n",
      "Epoch 14/100\n",
      "15/15 [==============================] - 0s 2ms/step - loss: 0.5288 - acc: 0.7829 - val_loss: 0.4929 - val_acc: 0.7984\n",
      "Epoch 15/100\n",
      "15/15 [==============================] - 0s 3ms/step - loss: 0.5023 - acc: 0.7995 - val_loss: 0.4901 - val_acc: 0.7984\n",
      "Epoch 16/100\n",
      "15/15 [==============================] - 0s 2ms/step - loss: 0.5042 - acc: 0.7904 - val_loss: 0.4852 - val_acc: 0.8000\n",
      "Epoch 17/100\n",
      "15/15 [==============================] - 0s 3ms/step - loss: 0.4933 - acc: 0.8059 - val_loss: 0.4821 - val_acc: 0.7920\n",
      "Epoch 18/100\n",
      "15/15 [==============================] - 0s 3ms/step - loss: 0.4668 - acc: 0.8187 - val_loss: 0.4776 - val_acc: 0.7952\n",
      "Epoch 19/100\n",
      "15/15 [==============================] - 0s 3ms/step - loss: 0.4895 - acc: 0.7957 - val_loss: 0.4708 - val_acc: 0.8000\n",
      "Epoch 20/100\n",
      "15/15 [==============================] - 0s 3ms/step - loss: 0.4587 - acc: 0.8139 - val_loss: 0.4691 - val_acc: 0.7984\n",
      "Epoch 21/100\n",
      "15/15 [==============================] - 0s 3ms/step - loss: 0.4618 - acc: 0.8069 - val_loss: 0.4665 - val_acc: 0.8000\n",
      "Epoch 22/100\n",
      "15/15 [==============================] - 0s 3ms/step - loss: 0.4502 - acc: 0.8048 - val_loss: 0.4632 - val_acc: 0.8000\n",
      "Epoch 23/100\n",
      "15/15 [==============================] - 0s 3ms/step - loss: 0.4418 - acc: 0.8208 - val_loss: 0.4649 - val_acc: 0.7952\n",
      "Epoch 24/100\n",
      "15/15 [==============================] - 0s 3ms/step - loss: 0.4535 - acc: 0.8171 - val_loss: 0.4603 - val_acc: 0.8000\n",
      "Epoch 25/100\n",
      "15/15 [==============================] - 0s 3ms/step - loss: 0.4231 - acc: 0.8299 - val_loss: 0.4568 - val_acc: 0.8032\n",
      "Epoch 26/100\n",
      "15/15 [==============================] - 0s 3ms/step - loss: 0.4166 - acc: 0.8208 - val_loss: 0.4581 - val_acc: 0.8080\n",
      "Epoch 27/100\n",
      "15/15 [==============================] - 0s 3ms/step - loss: 0.4210 - acc: 0.8187 - val_loss: 0.4567 - val_acc: 0.8000\n",
      "Epoch 28/100\n",
      "15/15 [==============================] - 0s 3ms/step - loss: 0.4202 - acc: 0.8315 - val_loss: 0.4504 - val_acc: 0.8032\n",
      "Epoch 29/100\n",
      "15/15 [==============================] - 0s 3ms/step - loss: 0.4148 - acc: 0.8299 - val_loss: 0.4515 - val_acc: 0.8064\n",
      "Epoch 30/100\n",
      "15/15 [==============================] - 0s 3ms/step - loss: 0.3991 - acc: 0.8299 - val_loss: 0.4506 - val_acc: 0.8064\n",
      "Epoch 31/100\n",
      "15/15 [==============================] - 0s 3ms/step - loss: 0.3970 - acc: 0.8347 - val_loss: 0.4500 - val_acc: 0.8048\n",
      "Epoch 32/100\n",
      "15/15 [==============================] - 0s 3ms/step - loss: 0.3955 - acc: 0.8363 - val_loss: 0.4531 - val_acc: 0.8032\n",
      "Epoch 33/100\n",
      "15/15 [==============================] - 0s 3ms/step - loss: 0.4023 - acc: 0.8304 - val_loss: 0.4464 - val_acc: 0.8128\n",
      "Epoch 34/100\n",
      "15/15 [==============================] - 0s 3ms/step - loss: 0.3947 - acc: 0.8411 - val_loss: 0.4479 - val_acc: 0.8112\n",
      "Epoch 35/100\n",
      "15/15 [==============================] - 0s 2ms/step - loss: 0.3857 - acc: 0.8421 - val_loss: 0.4447 - val_acc: 0.8032\n",
      "Epoch 36/100\n",
      "15/15 [==============================] - 0s 3ms/step - loss: 0.3790 - acc: 0.8496 - val_loss: 0.4432 - val_acc: 0.8144\n",
      "Epoch 37/100\n",
      "15/15 [==============================] - 0s 3ms/step - loss: 0.3782 - acc: 0.8416 - val_loss: 0.4404 - val_acc: 0.8096\n",
      "Epoch 38/100\n",
      "15/15 [==============================] - 0s 3ms/step - loss: 0.3734 - acc: 0.8469 - val_loss: 0.4413 - val_acc: 0.8112\n",
      "Epoch 39/100\n",
      "15/15 [==============================] - 0s 3ms/step - loss: 0.3733 - acc: 0.8475 - val_loss: 0.4399 - val_acc: 0.8096\n",
      "Epoch 40/100\n",
      "15/15 [==============================] - 0s 3ms/step - loss: 0.3708 - acc: 0.8459 - val_loss: 0.4407 - val_acc: 0.8112\n",
      "Epoch 41/100\n",
      "15/15 [==============================] - 0s 3ms/step - loss: 0.3557 - acc: 0.8544 - val_loss: 0.4359 - val_acc: 0.8176\n",
      "Epoch 42/100\n",
      "15/15 [==============================] - 0s 3ms/step - loss: 0.3666 - acc: 0.8411 - val_loss: 0.4379 - val_acc: 0.8176\n",
      "Epoch 43/100\n",
      "15/15 [==============================] - 0s 2ms/step - loss: 0.3592 - acc: 0.8560 - val_loss: 0.4389 - val_acc: 0.8128\n",
      "Epoch 44/100\n",
      "15/15 [==============================] - 0s 3ms/step - loss: 0.3522 - acc: 0.8528 - val_loss: 0.4368 - val_acc: 0.8160\n",
      "Epoch 45/100\n",
      "15/15 [==============================] - 0s 3ms/step - loss: 0.3600 - acc: 0.8480 - val_loss: 0.4358 - val_acc: 0.8176\n",
      "Epoch 46/100\n",
      "15/15 [==============================] - 0s 3ms/step - loss: 0.3467 - acc: 0.8613 - val_loss: 0.4359 - val_acc: 0.8160\n",
      "Epoch 47/100\n",
      "15/15 [==============================] - 0s 3ms/step - loss: 0.3504 - acc: 0.8555 - val_loss: 0.4397 - val_acc: 0.8208\n",
      "Epoch 48/100\n",
      "15/15 [==============================] - 0s 2ms/step - loss: 0.3340 - acc: 0.8677 - val_loss: 0.4378 - val_acc: 0.8176\n",
      "Epoch 49/100\n",
      "15/15 [==============================] - 0s 3ms/step - loss: 0.3433 - acc: 0.8539 - val_loss: 0.4391 - val_acc: 0.8192\n",
      "Epoch 50/100\n",
      "15/15 [==============================] - 0s 3ms/step - loss: 0.3364 - acc: 0.8560 - val_loss: 0.4409 - val_acc: 0.8160\n",
      "Epoch 51/100\n",
      "15/15 [==============================] - 0s 3ms/step - loss: 0.3323 - acc: 0.8635 - val_loss: 0.4426 - val_acc: 0.8192\n",
      "Epoch 52/100\n",
      "15/15 [==============================] - 0s 3ms/step - loss: 0.3326 - acc: 0.8624 - val_loss: 0.4434 - val_acc: 0.8176\n",
      "Epoch 53/100\n",
      "15/15 [==============================] - 0s 3ms/step - loss: 0.3317 - acc: 0.8677 - val_loss: 0.4445 - val_acc: 0.8096\n",
      "Epoch 54/100\n",
      "15/15 [==============================] - 0s 2ms/step - loss: 0.3358 - acc: 0.8603 - val_loss: 0.4429 - val_acc: 0.8176\n",
      "Epoch 55/100\n",
      "15/15 [==============================] - 0s 3ms/step - loss: 0.3245 - acc: 0.8656 - val_loss: 0.4437 - val_acc: 0.8176\n",
      "Epoch 56/100\n",
      "15/15 [==============================] - 0s 3ms/step - loss: 0.3321 - acc: 0.8555 - val_loss: 0.4425 - val_acc: 0.8128\n",
      "Epoch 57/100\n",
      "15/15 [==============================] - 0s 3ms/step - loss: 0.3247 - acc: 0.8608 - val_loss: 0.4402 - val_acc: 0.8192\n",
      "Epoch 58/100\n",
      "15/15 [==============================] - 0s 3ms/step - loss: 0.3262 - acc: 0.8592 - val_loss: 0.4419 - val_acc: 0.8192\n",
      "Epoch 59/100\n",
      "15/15 [==============================] - 0s 3ms/step - loss: 0.3099 - acc: 0.8683 - val_loss: 0.4389 - val_acc: 0.8144\n",
      "Epoch 60/100\n",
      "15/15 [==============================] - 0s 2ms/step - loss: 0.3255 - acc: 0.8629 - val_loss: 0.4375 - val_acc: 0.8176\n",
      "Epoch 61/100\n",
      "15/15 [==============================] - 0s 3ms/step - loss: 0.3101 - acc: 0.8715 - val_loss: 0.4358 - val_acc: 0.8256\n",
      "Epoch 62/100\n",
      "15/15 [==============================] - 0s 3ms/step - loss: 0.3059 - acc: 0.8688 - val_loss: 0.4309 - val_acc: 0.8192\n",
      "Epoch 63/100\n",
      "15/15 [==============================] - 0s 3ms/step - loss: 0.3083 - acc: 0.8656 - val_loss: 0.4332 - val_acc: 0.8208\n",
      "Epoch 64/100\n",
      "15/15 [==============================] - 0s 2ms/step - loss: 0.3011 - acc: 0.8747 - val_loss: 0.4396 - val_acc: 0.8160\n",
      "Epoch 65/100\n",
      "15/15 [==============================] - 0s 3ms/step - loss: 0.3020 - acc: 0.8731 - val_loss: 0.4463 - val_acc: 0.8240\n",
      "Epoch 66/100\n",
      "15/15 [==============================] - 0s 3ms/step - loss: 0.3142 - acc: 0.8683 - val_loss: 0.4401 - val_acc: 0.8272\n",
      "Epoch 67/100\n",
      "15/15 [==============================] - 0s 3ms/step - loss: 0.2991 - acc: 0.8720 - val_loss: 0.4368 - val_acc: 0.8272\n",
      "Epoch 68/100\n",
      "15/15 [==============================] - 0s 3ms/step - loss: 0.3082 - acc: 0.8715 - val_loss: 0.4454 - val_acc: 0.8176\n",
      "Epoch 69/100\n",
      "15/15 [==============================] - 0s 3ms/step - loss: 0.3065 - acc: 0.8661 - val_loss: 0.4451 - val_acc: 0.8208\n",
      "Epoch 70/100\n",
      "15/15 [==============================] - 0s 3ms/step - loss: 0.2927 - acc: 0.8699 - val_loss: 0.4464 - val_acc: 0.8208\n",
      "Epoch 71/100\n",
      "15/15 [==============================] - 0s 3ms/step - loss: 0.2947 - acc: 0.8725 - val_loss: 0.4452 - val_acc: 0.8224\n",
      "Epoch 72/100\n",
      "15/15 [==============================] - 0s 3ms/step - loss: 0.2939 - acc: 0.8779 - val_loss: 0.4427 - val_acc: 0.8192\n",
      "Epoch 73/100\n",
      "15/15 [==============================] - 0s 3ms/step - loss: 0.2906 - acc: 0.8763 - val_loss: 0.4405 - val_acc: 0.8192\n",
      "Epoch 74/100\n",
      "15/15 [==============================] - 0s 3ms/step - loss: 0.2830 - acc: 0.8811 - val_loss: 0.4480 - val_acc: 0.8208\n",
      "Epoch 75/100\n",
      "15/15 [==============================] - 0s 3ms/step - loss: 0.2928 - acc: 0.8789 - val_loss: 0.4487 - val_acc: 0.8160\n",
      "Epoch 76/100\n",
      "15/15 [==============================] - 0s 3ms/step - loss: 0.2747 - acc: 0.8832 - val_loss: 0.4540 - val_acc: 0.8192\n",
      "Epoch 77/100\n",
      "15/15 [==============================] - 0s 3ms/step - loss: 0.2801 - acc: 0.8779 - val_loss: 0.4557 - val_acc: 0.8160\n",
      "Epoch 78/100\n",
      "15/15 [==============================] - 0s 3ms/step - loss: 0.2836 - acc: 0.8784 - val_loss: 0.4484 - val_acc: 0.8256\n",
      "Epoch 79/100\n",
      "15/15 [==============================] - 0s 3ms/step - loss: 0.2773 - acc: 0.8773 - val_loss: 0.4542 - val_acc: 0.8240\n",
      "Epoch 80/100\n",
      "15/15 [==============================] - 0s 3ms/step - loss: 0.2791 - acc: 0.8811 - val_loss: 0.4485 - val_acc: 0.8208\n",
      "Epoch 81/100\n",
      "15/15 [==============================] - 0s 2ms/step - loss: 0.2723 - acc: 0.8763 - val_loss: 0.4536 - val_acc: 0.8208\n",
      "Epoch 82/100\n",
      "15/15 [==============================] - 0s 2ms/step - loss: 0.2727 - acc: 0.8811 - val_loss: 0.4593 - val_acc: 0.8256\n",
      "Epoch 83/100\n",
      "15/15 [==============================] - 0s 2ms/step - loss: 0.2741 - acc: 0.8848 - val_loss: 0.4602 - val_acc: 0.8192\n",
      "Epoch 84/100\n",
      "15/15 [==============================] - 0s 3ms/step - loss: 0.2734 - acc: 0.8800 - val_loss: 0.4641 - val_acc: 0.8224\n",
      "Epoch 85/100\n",
      "15/15 [==============================] - 0s 3ms/step - loss: 0.2656 - acc: 0.8725 - val_loss: 0.4654 - val_acc: 0.8208\n",
      "Epoch 86/100\n",
      "15/15 [==============================] - 0s 2ms/step - loss: 0.2623 - acc: 0.8821 - val_loss: 0.4636 - val_acc: 0.8224\n",
      "Epoch 87/100\n",
      "15/15 [==============================] - 0s 3ms/step - loss: 0.2712 - acc: 0.8779 - val_loss: 0.4647 - val_acc: 0.8176\n",
      "Epoch 88/100\n",
      "15/15 [==============================] - 0s 3ms/step - loss: 0.2696 - acc: 0.8891 - val_loss: 0.4686 - val_acc: 0.8192\n",
      "Epoch 89/100\n",
      "15/15 [==============================] - 0s 3ms/step - loss: 0.2638 - acc: 0.8773 - val_loss: 0.4667 - val_acc: 0.8288\n",
      "Epoch 90/100\n",
      "15/15 [==============================] - 0s 3ms/step - loss: 0.2743 - acc: 0.8773 - val_loss: 0.4644 - val_acc: 0.8192\n",
      "Epoch 91/100\n",
      "15/15 [==============================] - 0s 3ms/step - loss: 0.2637 - acc: 0.8832 - val_loss: 0.4596 - val_acc: 0.8192\n",
      "Epoch 92/100\n",
      "15/15 [==============================] - 0s 3ms/step - loss: 0.2598 - acc: 0.8917 - val_loss: 0.4709 - val_acc: 0.8240\n",
      "Epoch 93/100\n",
      "15/15 [==============================] - 0s 3ms/step - loss: 0.2593 - acc: 0.8848 - val_loss: 0.4722 - val_acc: 0.8176\n",
      "Epoch 94/100\n",
      "15/15 [==============================] - 0s 3ms/step - loss: 0.2645 - acc: 0.8827 - val_loss: 0.4720 - val_acc: 0.8256\n",
      "Epoch 95/100\n",
      "15/15 [==============================] - 0s 2ms/step - loss: 0.2520 - acc: 0.8869 - val_loss: 0.4804 - val_acc: 0.8160\n",
      "Epoch 96/100\n",
      "15/15 [==============================] - 0s 3ms/step - loss: 0.2567 - acc: 0.8795 - val_loss: 0.4789 - val_acc: 0.8240\n",
      "Epoch 97/100\n",
      "15/15 [==============================] - 0s 3ms/step - loss: 0.2580 - acc: 0.8875 - val_loss: 0.4780 - val_acc: 0.8160\n",
      "Epoch 98/100\n",
      "15/15 [==============================] - 0s 3ms/step - loss: 0.2585 - acc: 0.8875 - val_loss: 0.4772 - val_acc: 0.8208\n",
      "Epoch 99/100\n",
      "15/15 [==============================] - 0s 2ms/step - loss: 0.2521 - acc: 0.8907 - val_loss: 0.4864 - val_acc: 0.8176\n",
      "Epoch 100/100\n",
      "15/15 [==============================] - 0s 2ms/step - loss: 0.2531 - acc: 0.8795 - val_loss: 0.4885 - val_acc: 0.8224\n"
     ]
    }
   ],
   "source": [
    "fnn_model = Sequential()\n",
    "\n",
    "fnn_model.add(Dense(16, activation='relu', input_shape=(X_train_sc.shape[1],)))\n",
    "fnn_model.add(Dense(64, activation='relu'))\n",
    "fnn_model.add(Dropout(0.2))\n",
    "fnn_model.add(Dense(32, activation='relu'))\n",
    "fnn_model.add(Dropout(0.2))\n",
    "fnn_model.add(Dense(16, activation='relu'))\n",
    "fnn_model.add(Dense(5, activation='softmax'))\n",
    "\n",
    "fnn_model.compile(optimizer='Adam', loss='categorical_crossentropy', metrics=['acc'])\n",
    "\n",
    "hist = fnn_model.fit(X_train_sc, y_train,\n",
    "                epochs=100,\n",
    "                batch_size=128,\n",
    "                validation_data=(X_test_sc, y_test),\n",
    "                    use_multiprocessing=True,\n",
    "                    verbose=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 300,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(hist.history['acc'], label=\"Training Acc\")\n",
    "plt.plot(hist.history['val_acc'], label=\"Testing Acc\")\n",
    "plt.legend();"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "With my extracted features (energy, rms, and mfcc), I'm able to improve on the accuracy of the spotify model by **5-7% on training data**\n",
    "* SVC model $\\approx 84\\%$ accuracy\n",
    "* FNN model $\\approx 82\\%$ accuracy"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## What's next?"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "I still want to find some way for the model to be able to tell the difference between rap and r&b"
   ]
  }
 ],
 "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": "3.7.9"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}