MLFinal.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/opt/homebrew/lib/python3.11/site-packages/pandas/core/arraylike.py:396: RuntimeWarning: invalid value encountered in log1p\n",
      "  result = getattr(ufunc, method)(*inputs, **kwargs)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "Index: 17170 entries, 0 to 17169\n",
      "Data columns (total 32 columns):\n",
      " #   Column                  Non-Null Count  Dtype  \n",
      "---  ------                  --------------  -----  \n",
      " 0   Danceability            17170 non-null  float64\n",
      " 1   Energy                  14615 non-null  float64\n",
      " 2   Loudness                14601 non-null  float64\n",
      " 3   Speechiness             14655 non-null  float64\n",
      " 4   Acousticness            14593 non-null  float64\n",
      " 5   Instrumentalness        14569 non-null  float64\n",
      " 6   Liveness                14635 non-null  float64\n",
      " 7   Valence                 14596 non-null  float64\n",
      " 8   Tempo                   14608 non-null  float64\n",
      " 9   Duration_ms             14567 non-null  float64\n",
      " 10  Views                   14574 non-null  float64\n",
      " 11  Likes                   14591 non-null  float64\n",
      " 12  Stream                  14602 non-null  float64\n",
      " 13  Comments                14589 non-null  float64\n",
      " 14  Album_type_album        17170 non-null  bool   \n",
      " 15  Album_type_compilation  17170 non-null  bool   \n",
      " 16  Album_type_single       17170 non-null  bool   \n",
      " 17  Licensed_False          17170 non-null  bool   \n",
      " 18  Licensed_True           17170 non-null  bool   \n",
      " 19  official_video_False    17170 non-null  bool   \n",
      " 20  official_video_True     17170 non-null  bool   \n",
      " 21  Key_0.0                 17170 non-null  bool   \n",
      " 22  Key_1.0                 17170 non-null  bool   \n",
      " 23  Key_2.0                 17170 non-null  bool   \n",
      " 24  Key_3.0                 17170 non-null  bool   \n",
      " 25  Key_4.0                 17170 non-null  bool   \n",
      " 26  Key_5.0                 17170 non-null  bool   \n",
      " 27  Key_6.0                 17170 non-null  bool   \n",
      " 28  Key_7.0                 17170 non-null  bool   \n",
      " 29  Key_8.0                 17170 non-null  bool   \n",
      " 30  Key_9.0                 17170 non-null  bool   \n",
      " 31  Key_10.0                17170 non-null  bool   \n",
      "dtypes: bool(18), float64(14)\n",
      "memory usage: 2.3 MB\n",
      "<class 'pandas.core.frame.DataFrame'>\n",
      "Index: 6315 entries, 17170 to 23484\n",
      "Data columns (total 31 columns):\n",
      " #   Column                  Non-Null Count  Dtype  \n",
      "---  ------                  --------------  -----  \n",
      " 0   Energy                  5347 non-null   float64\n",
      " 1   Loudness                5361 non-null   float64\n",
      " 2   Speechiness             5307 non-null   float64\n",
      " 3   Acousticness            5369 non-null   float64\n",
      " 4   Instrumentalness        5393 non-null   float64\n",
      " 5   Liveness                5327 non-null   float64\n",
      " 6   Valence                 5366 non-null   float64\n",
      " 7   Tempo                   5354 non-null   float64\n",
      " 8   Duration_ms             5257 non-null   float64\n",
      " 9   Views                   4533 non-null   float64\n",
      " 10  Likes                   4461 non-null   float64\n",
      " 11  Stream                  4774 non-null   float64\n",
      " 12  Comments                5373 non-null   float64\n",
      " 13  Album_type_album        6315 non-null   bool   \n",
      " 14  Album_type_compilation  6315 non-null   bool   \n",
      " 15  Album_type_single       6315 non-null   bool   \n",
      " 16  Licensed_False          6315 non-null   bool   \n",
      " 17  Licensed_True           6315 non-null   bool   \n",
      " 18  official_video_False    6315 non-null   bool   \n",
      " 19  official_video_True     6315 non-null   bool   \n",
      " 20  Key_0.0                 6315 non-null   bool   \n",
      " 21  Key_1.0                 6315 non-null   bool   \n",
      " 22  Key_2.0                 6315 non-null   bool   \n",
      " 23  Key_3.0                 6315 non-null   bool   \n",
      " 24  Key_4.0                 6315 non-null   bool   \n",
      " 25  Key_5.0                 6315 non-null   bool   \n",
      " 26  Key_6.0                 6315 non-null   bool   \n",
      " 27  Key_7.0                 6315 non-null   bool   \n",
      " 28  Key_8.0                 6315 non-null   bool   \n",
      " 29  Key_9.0                 6315 non-null   bool   \n",
      " 30  Key_10.0                6315 non-null   bool   \n",
      "dtypes: bool(18), float64(13)\n",
      "memory usage: 801.7 KB\n",
      "(17170, 528)\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Energy</th>\n",
       "      <th>Loudness</th>\n",
       "      <th>Speechiness</th>\n",
       "      <th>Acousticness</th>\n",
       "      <th>Instrumentalness</th>\n",
       "      <th>Liveness</th>\n",
       "      <th>Valence</th>\n",
       "      <th>Tempo</th>\n",
       "      <th>Duration_ms</th>\n",
       "      <th>Views</th>\n",
       "      <th>...</th>\n",
       "      <th>Key_1.0</th>\n",
       "      <th>Key_2.0</th>\n",
       "      <th>Key_3.0</th>\n",
       "      <th>Key_4.0</th>\n",
       "      <th>Key_5.0</th>\n",
       "      <th>Key_6.0</th>\n",
       "      <th>Key_7.0</th>\n",
       "      <th>Key_8.0</th>\n",
       "      <th>Key_9.0</th>\n",
       "      <th>Key_10.0</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>id</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>17170</th>\n",
       "      <td>0.350403</td>\n",
       "      <td>-6.679000</td>\n",
       "      <td>0.177000</td>\n",
       "      <td>5.842771e-07</td>\n",
       "      <td>0.002330</td>\n",
       "      <td>0.230346</td>\n",
       "      <td>0.488375</td>\n",
       "      <td>138.559000</td>\n",
       "      <td>12.313316</td>\n",
       "      <td>20.357341</td>\n",
       "      <td>...</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>17171</th>\n",
       "      <td>0.347429</td>\n",
       "      <td>-5.815000</td>\n",
       "      <td>0.030200</td>\n",
       "      <td>1.531935e-01</td>\n",
       "      <td>0.000687</td>\n",
       "      <td>0.000099</td>\n",
       "      <td>0.852000</td>\n",
       "      <td>92.761000</td>\n",
       "      <td>12.206942</td>\n",
       "      <td>18.092338</td>\n",
       "      <td>...</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>17172</th>\n",
       "      <td>0.786330</td>\n",
       "      <td>-3.930000</td>\n",
       "      <td>0.052200</td>\n",
       "      <td>7.676563e-05</td>\n",
       "      <td>0.114404</td>\n",
       "      <td>0.001561</td>\n",
       "      <td>0.551000</td>\n",
       "      <td>108.014000</td>\n",
       "      <td>12.279095</td>\n",
       "      <td>15.947907</td>\n",
       "      <td>...</td>\n",
       "      <td>True</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>17173</th>\n",
       "      <td>0.403583</td>\n",
       "      <td>-5.810000</td>\n",
       "      <td>0.026000</td>\n",
       "      <td>3.442951e-15</td>\n",
       "      <td>0.509000</td>\n",
       "      <td>0.000262</td>\n",
       "      <td>0.578000</td>\n",
       "      <td>120.423000</td>\n",
       "      <td>12.362512</td>\n",
       "      <td>16.833847</td>\n",
       "      <td>...</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>17174</th>\n",
       "      <td>0.334255</td>\n",
       "      <td>-8.627000</td>\n",
       "      <td>0.171000</td>\n",
       "      <td>1.619428e-05</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000340</td>\n",
       "      <td>0.525000</td>\n",
       "      <td>167.953000</td>\n",
       "      <td>12.739406</td>\n",
       "      <td>20.242777</td>\n",
       "      <td>...</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>23480</th>\n",
       "      <td>0.160377</td>\n",
       "      <td>-6.366894</td>\n",
       "      <td>0.045451</td>\n",
       "      <td>3.176921e-01</td>\n",
       "      <td>0.006016</td>\n",
       "      <td>0.001642</td>\n",
       "      <td>0.160253</td>\n",
       "      <td>72.136829</td>\n",
       "      <td>12.361892</td>\n",
       "      <td>16.081098</td>\n",
       "      <td>...</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>23481</th>\n",
       "      <td>0.318585</td>\n",
       "      <td>-4.148566</td>\n",
       "      <td>0.073000</td>\n",
       "      <td>7.419596e-03</td>\n",
       "      <td>0.000817</td>\n",
       "      <td>0.008843</td>\n",
       "      <td>0.614366</td>\n",
       "      <td>114.903993</td>\n",
       "      <td>11.840266</td>\n",
       "      <td>17.638231</td>\n",
       "      <td>...</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>23482</th>\n",
       "      <td>0.426223</td>\n",
       "      <td>-5.807545</td>\n",
       "      <td>0.072906</td>\n",
       "      <td>6.617082e-01</td>\n",
       "      <td>0.898308</td>\n",
       "      <td>0.000662</td>\n",
       "      <td>0.087281</td>\n",
       "      <td>87.969239</td>\n",
       "      <td>12.132658</td>\n",
       "      <td>16.833847</td>\n",
       "      <td>...</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>23483</th>\n",
       "      <td>0.204440</td>\n",
       "      <td>-11.563964</td>\n",
       "      <td>0.037676</td>\n",
       "      <td>3.630868e-03</td>\n",
       "      <td>0.491785</td>\n",
       "      <td>0.030605</td>\n",
       "      <td>0.482987</td>\n",
       "      <td>114.903993</td>\n",
       "      <td>12.104263</td>\n",
       "      <td>16.665138</td>\n",
       "      <td>...</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>23484</th>\n",
       "      <td>0.014679</td>\n",
       "      <td>-8.434696</td>\n",
       "      <td>0.082686</td>\n",
       "      <td>7.642603e-02</td>\n",
       "      <td>0.214188</td>\n",
       "      <td>0.001583</td>\n",
       "      <td>0.735073</td>\n",
       "      <td>98.873933</td>\n",
       "      <td>11.985627</td>\n",
       "      <td>16.833847</td>\n",
       "      <td>...</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>6315 rows × 31 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "         Energy   Loudness  Speechiness  Acousticness  Instrumentalness   \n",
       "id                                                                        \n",
       "17170  0.350403  -6.679000     0.177000  5.842771e-07          0.002330  \\\n",
       "17171  0.347429  -5.815000     0.030200  1.531935e-01          0.000687   \n",
       "17172  0.786330  -3.930000     0.052200  7.676563e-05          0.114404   \n",
       "17173  0.403583  -5.810000     0.026000  3.442951e-15          0.509000   \n",
       "17174  0.334255  -8.627000     0.171000  1.619428e-05          0.000000   \n",
       "...         ...        ...          ...           ...               ...   \n",
       "23480  0.160377  -6.366894     0.045451  3.176921e-01          0.006016   \n",
       "23481  0.318585  -4.148566     0.073000  7.419596e-03          0.000817   \n",
       "23482  0.426223  -5.807545     0.072906  6.617082e-01          0.898308   \n",
       "23483  0.204440 -11.563964     0.037676  3.630868e-03          0.491785   \n",
       "23484  0.014679  -8.434696     0.082686  7.642603e-02          0.214188   \n",
       "\n",
       "       Liveness   Valence       Tempo  Duration_ms      Views  ...  Key_1.0   \n",
       "id                                                             ...            \n",
       "17170  0.230346  0.488375  138.559000    12.313316  20.357341  ...    False  \\\n",
       "17171  0.000099  0.852000   92.761000    12.206942  18.092338  ...    False   \n",
       "17172  0.001561  0.551000  108.014000    12.279095  15.947907  ...     True   \n",
       "17173  0.000262  0.578000  120.423000    12.362512  16.833847  ...    False   \n",
       "17174  0.000340  0.525000  167.953000    12.739406  20.242777  ...    False   \n",
       "...         ...       ...         ...          ...        ...  ...      ...   \n",
       "23480  0.001642  0.160253   72.136829    12.361892  16.081098  ...    False   \n",
       "23481  0.008843  0.614366  114.903993    11.840266  17.638231  ...    False   \n",
       "23482  0.000662  0.087281   87.969239    12.132658  16.833847  ...    False   \n",
       "23483  0.030605  0.482987  114.903993    12.104263  16.665138  ...    False   \n",
       "23484  0.001583  0.735073   98.873933    11.985627  16.833847  ...    False   \n",
       "\n",
       "       Key_2.0  Key_3.0  Key_4.0  Key_5.0  Key_6.0  Key_7.0  Key_8.0  Key_9.0   \n",
       "id                                                                              \n",
       "17170    False    False    False    False     True    False    False    False  \\\n",
       "17171    False    False    False    False    False    False     True    False   \n",
       "17172    False    False    False    False    False    False    False    False   \n",
       "17173     True    False    False    False    False    False    False    False   \n",
       "17174    False    False    False    False    False    False    False    False   \n",
       "...        ...      ...      ...      ...      ...      ...      ...      ...   \n",
       "23480    False    False    False    False    False     True    False    False   \n",
       "23481    False    False    False    False    False    False    False    False   \n",
       "23482    False    False    False     True    False    False    False    False   \n",
       "23483    False    False    False    False    False    False    False    False   \n",
       "23484    False    False    False    False    False    False    False    False   \n",
       "\n",
       "       Key_10.0  \n",
       "id               \n",
       "17170     False  \n",
       "17171     False  \n",
       "17172     False  \n",
       "17173     False  \n",
       "17174     False  \n",
       "...         ...  \n",
       "23480     False  \n",
       "23481      True  \n",
       "23482     False  \n",
       "23483     False  \n",
       "23484     False  \n",
       "\n",
       "[6315 rows x 31 columns]"
      ]
     },
     "execution_count": 1,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "import numpy as np\n",
    "import pandas as pd\n",
    "from tqdm import tqdm\n",
    "from sklearn.linear_model import *\n",
    "from sklearn.model_selection import *\n",
    "from sklearn.impute import *\n",
    "from sklearn.tree import *\n",
    "from sklearn.ensemble import *\n",
    "from sklearn.metrics import *\n",
    "from sklearn.preprocessing import *\n",
    "import matplotlib.pyplot as plt\n",
    "import xgboost as xgb\n",
    "\n",
    "\n",
    "#input datas\n",
    "df_train = pd.read_csv(\"train.csv\", index_col='id')\n",
    "df_test = pd.read_csv(\"test.csv\", index_col='id')\n",
    "\n",
    "for col in ['Views', 'Likes', 'Stream', 'Comments', 'Duration_ms']:\n",
    "    df_train[col] = np.log1p(df_train[col])\n",
    "    df_test[col] = np.log1p(df_test[col])\n",
    "    \n",
    "    \n",
    "df_train['Key'] = df_train['Key'].apply(lambda x: x if np.isnan(x) else int(x) )\n",
    "df_test['Key'] = df_test['Key'].apply(lambda x: x if np.isnan(x) else int(x) )\n",
    "\n",
    "for feature in ['Album_type', 'Licensed', 'official_video', 'Key']:\n",
    "    df_train = pd.get_dummies(df_train, columns=[feature], prefix=feature)\n",
    "    df_test = pd.get_dummies(df_test, columns=[feature], prefix=feature)\n",
    "\n",
    "df_train = df_train.select_dtypes(exclude=['object'])\n",
    "df_test = df_test.select_dtypes(exclude=['object'])\n",
    "\n",
    "df_train.info()\n",
    "df_test.info()\n",
    "df_test.shape\n",
    "\n",
    "#impute data\n",
    "\n",
    "#with KNN\n",
    "#imputer = KNNImputer(n_neighbors=5) \n",
    "#df_train = pd.DataFrame(imputer.fit_transform(df_train), columns = df_train.columns)\n",
    "#df_test = pd.DataFrame(imputer.fit_transform(df_test), columns = df_test.columns)\n",
    "\n",
    "#fill missing value in training set with mean\n",
    "df_train = df_train.fillna(df_train.mean())\n",
    "df_test = df_test.fillna(df_test.mean()) \n",
    "\n",
    "#df_train = df_train.fillna(df_train.median())\n",
    "\n",
    "#fill missing value in testing set with mean\n",
    "#df_test = df_test.fillna(df_test.median())\n",
    "\n",
    "#convert training and testing set to numpy array\n",
    "X_test = df_test.to_numpy()\n",
    "X = df_train.drop([\"Danceability\"], axis=1).to_numpy()\n",
    "Y = df_train['Danceability'].to_numpy()\n",
    "\n",
    "#use polynomial features\n",
    "\n",
    "poly = PolynomialFeatures(2)\n",
    "X = poly.fit_transform(X)\n",
    "X_test = poly.fit_transform(X_test)\n",
    "\n",
    "\n",
    "#normalize the training and testing data\n",
    "trainscaler = StandardScaler()\n",
    "trainscaler.fit(X)\n",
    "X = trainscaler.transform(X)\n",
    "testscaler = StandardScaler()\n",
    "testscaler.fit(X_test)\n",
    "X_test = testscaler.transform(X_test)\n",
    "print(X.shape)\n",
    "df_test"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "  0%|          | 0/9 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 9/9 [04:55<00:00, 32.80s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[1.706018632788867, 1.7030446425253871, 1.7112977839891699, 1.70130000157777, 1.702145097059032, 1.6826819549549543, 1.7057095171543069, 2.103893004771784, 2.4761793826441467]\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "#use cross validation to test the optimal coefficient of the L1-regularization term(alpha) in xgboost \n",
    "#with absolute error as the loss function\n",
    "kfold = KFold(n_splits=5, shuffle=True)\n",
    "alpha_value = [-4, -3, -2, -1, 0, 1, 2, 3, 4]\n",
    "maes = []\n",
    "\n",
    "for a in tqdm(alpha_value):\n",
    "    perform = []\n",
    "    for train_index, test_index in kfold.split(X):\n",
    "        train_x, valid_x = X[train_index], X[test_index]\n",
    "        train_y, valid_y = Y[train_index], Y[test_index]\n",
    "        model = xgb.XGBRegressor(objective=\"reg:absoluteerror\", alpha = 10**a, reg_lambda = 0,seed=123)\n",
    "        #labmda is the L2-regularization term, we set it to zero\n",
    "        model.fit(train_x, train_y)\n",
    "        pred_y = model.predict(valid_x)\n",
    "        perform.append(mean_absolute_error(valid_y, pred_y))\n",
    "    maes.append(np.mean(perform))    \n",
    "    \n",
    "print(maes)\n",
    "\n",
    "#plot the graph of alpha_values and maes\n",
    "plt.plot(alpha_value, maes)\n",
    "plt.xlabel('alpha_value')\n",
    "plt.ylabel('Mean Absolute Error')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "  0%|          | 0/9 [00:04<?, ?it/s]\n"
     ]
    },
    {
     "ename": "KeyboardInterrupt",
     "evalue": "",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[0;31mKeyboardInterrupt\u001b[0m                         Traceback (most recent call last)",
      "Cell \u001b[0;32mIn[26], line 13\u001b[0m\n\u001b[1;32m     11\u001b[0m train_y, valid_y \u001b[39m=\u001b[39m Y[train_index], Y[test_index]\n\u001b[1;32m     12\u001b[0m model \u001b[39m=\u001b[39m xgb\u001b[39m.\u001b[39mXGBRegressor(objective\u001b[39m=\u001b[39m\u001b[39m\"\u001b[39m\u001b[39mreg:pseudohubererror\u001b[39m\u001b[39m\"\u001b[39m, alpha \u001b[39m=\u001b[39m \u001b[39m10\u001b[39m\u001b[39m*\u001b[39m\u001b[39m*\u001b[39ma,reg_lambda \u001b[39m=\u001b[39m \u001b[39m0\u001b[39m,seed\u001b[39m=\u001b[39m\u001b[39m123\u001b[39m)\n\u001b[0;32m---> 13\u001b[0m model\u001b[39m.\u001b[39;49mfit(train_x, train_y)\n\u001b[1;32m     14\u001b[0m pred_y \u001b[39m=\u001b[39m model\u001b[39m.\u001b[39mpredict(valid_x)\n\u001b[1;32m     15\u001b[0m perform\u001b[39m.\u001b[39mappend(mean_absolute_error(valid_y, pred_y))\n",
      "File \u001b[0;32m/opt/homebrew/lib/python3.11/site-packages/xgboost/core.py:620\u001b[0m, in \u001b[0;36mrequire_keyword_args.<locals>.throw_if.<locals>.inner_f\u001b[0;34m(*args, **kwargs)\u001b[0m\n\u001b[1;32m    618\u001b[0m \u001b[39mfor\u001b[39;00m k, arg \u001b[39min\u001b[39;00m \u001b[39mzip\u001b[39m(sig\u001b[39m.\u001b[39mparameters, args):\n\u001b[1;32m    619\u001b[0m     kwargs[k] \u001b[39m=\u001b[39m arg\n\u001b[0;32m--> 620\u001b[0m \u001b[39mreturn\u001b[39;00m func(\u001b[39m*\u001b[39;49m\u001b[39m*\u001b[39;49mkwargs)\n",
      "File \u001b[0;32m/opt/homebrew/lib/python3.11/site-packages/xgboost/sklearn.py:1025\u001b[0m, in \u001b[0;36mXGBModel.fit\u001b[0;34m(self, X, y, sample_weight, base_margin, eval_set, eval_metric, early_stopping_rounds, verbose, xgb_model, sample_weight_eval_set, base_margin_eval_set, feature_weights, callbacks)\u001b[0m\n\u001b[1;32m   1014\u001b[0m     obj \u001b[39m=\u001b[39m \u001b[39mNone\u001b[39;00m\n\u001b[1;32m   1016\u001b[0m (\n\u001b[1;32m   1017\u001b[0m     model,\n\u001b[1;32m   1018\u001b[0m     metric,\n\u001b[0;32m   (...)\u001b[0m\n\u001b[1;32m   1023\u001b[0m     xgb_model, eval_metric, params, early_stopping_rounds, callbacks\n\u001b[1;32m   1024\u001b[0m )\n\u001b[0;32m-> 1025\u001b[0m \u001b[39mself\u001b[39m\u001b[39m.\u001b[39m_Booster \u001b[39m=\u001b[39m train(\n\u001b[1;32m   1026\u001b[0m     params,\n\u001b[1;32m   1027\u001b[0m     train_dmatrix,\n\u001b[1;32m   1028\u001b[0m     \u001b[39mself\u001b[39;49m\u001b[39m.\u001b[39;49mget_num_boosting_rounds(),\n\u001b[1;32m   1029\u001b[0m     evals\u001b[39m=\u001b[39;49mevals,\n\u001b[1;32m   1030\u001b[0m     early_stopping_rounds\u001b[39m=\u001b[39;49mearly_stopping_rounds,\n\u001b[1;32m   1031\u001b[0m     evals_result\u001b[39m=\u001b[39;49mevals_result,\n\u001b[1;32m   1032\u001b[0m     obj\u001b[39m=\u001b[39;49mobj,\n\u001b[1;32m   1033\u001b[0m     custom_metric\u001b[39m=\u001b[39;49mmetric,\n\u001b[1;32m   1034\u001b[0m     verbose_eval\u001b[39m=\u001b[39;49mverbose,\n\u001b[1;32m   1035\u001b[0m     xgb_model\u001b[39m=\u001b[39;49mmodel,\n\u001b[1;32m   1036\u001b[0m     callbacks\u001b[39m=\u001b[39;49mcallbacks,\n\u001b[1;32m   1037\u001b[0m )\n\u001b[1;32m   1039\u001b[0m \u001b[39mself\u001b[39m\u001b[39m.\u001b[39m_set_evaluation_result(evals_result)\n\u001b[1;32m   1040\u001b[0m \u001b[39mreturn\u001b[39;00m \u001b[39mself\u001b[39m\n",
      "File \u001b[0;32m/opt/homebrew/lib/python3.11/site-packages/xgboost/core.py:620\u001b[0m, in \u001b[0;36mrequire_keyword_args.<locals>.throw_if.<locals>.inner_f\u001b[0;34m(*args, **kwargs)\u001b[0m\n\u001b[1;32m    618\u001b[0m \u001b[39mfor\u001b[39;00m k, arg \u001b[39min\u001b[39;00m \u001b[39mzip\u001b[39m(sig\u001b[39m.\u001b[39mparameters, args):\n\u001b[1;32m    619\u001b[0m     kwargs[k] \u001b[39m=\u001b[39m arg\n\u001b[0;32m--> 620\u001b[0m \u001b[39mreturn\u001b[39;00m func(\u001b[39m*\u001b[39;49m\u001b[39m*\u001b[39;49mkwargs)\n",
      "File \u001b[0;32m/opt/homebrew/lib/python3.11/site-packages/xgboost/training.py:185\u001b[0m, in \u001b[0;36mtrain\u001b[0;34m(params, dtrain, num_boost_round, evals, obj, feval, maximize, early_stopping_rounds, evals_result, verbose_eval, xgb_model, callbacks, custom_metric)\u001b[0m\n\u001b[1;32m    183\u001b[0m \u001b[39mif\u001b[39;00m cb_container\u001b[39m.\u001b[39mbefore_iteration(bst, i, dtrain, evals):\n\u001b[1;32m    184\u001b[0m     \u001b[39mbreak\u001b[39;00m\n\u001b[0;32m--> 185\u001b[0m bst\u001b[39m.\u001b[39;49mupdate(dtrain, i, obj)\n\u001b[1;32m    186\u001b[0m \u001b[39mif\u001b[39;00m cb_container\u001b[39m.\u001b[39mafter_iteration(bst, i, dtrain, evals):\n\u001b[1;32m    187\u001b[0m     \u001b[39mbreak\u001b[39;00m\n",
      "File \u001b[0;32m/opt/homebrew/lib/python3.11/site-packages/xgboost/core.py:1918\u001b[0m, in \u001b[0;36mBooster.update\u001b[0;34m(self, dtrain, iteration, fobj)\u001b[0m\n\u001b[1;32m   1915\u001b[0m \u001b[39mself\u001b[39m\u001b[39m.\u001b[39m_validate_dmatrix_features(dtrain)\n\u001b[1;32m   1917\u001b[0m \u001b[39mif\u001b[39;00m fobj \u001b[39mis\u001b[39;00m \u001b[39mNone\u001b[39;00m:\n\u001b[0;32m-> 1918\u001b[0m     _check_call(_LIB\u001b[39m.\u001b[39;49mXGBoosterUpdateOneIter(\u001b[39mself\u001b[39;49m\u001b[39m.\u001b[39;49mhandle,\n\u001b[1;32m   1919\u001b[0m                                             ctypes\u001b[39m.\u001b[39;49mc_int(iteration),\n\u001b[1;32m   1920\u001b[0m                                             dtrain\u001b[39m.\u001b[39;49mhandle))\n\u001b[1;32m   1921\u001b[0m \u001b[39melse\u001b[39;00m:\n\u001b[1;32m   1922\u001b[0m     pred \u001b[39m=\u001b[39m \u001b[39mself\u001b[39m\u001b[39m.\u001b[39mpredict(dtrain, output_margin\u001b[39m=\u001b[39m\u001b[39mTrue\u001b[39;00m, training\u001b[39m=\u001b[39m\u001b[39mTrue\u001b[39;00m)\n",
      "\u001b[0;31mKeyboardInterrupt\u001b[0m: "
     ]
    }
   ],
   "source": [
    "#use cross validation to test the optimal coefficient of the regularization term(alpha) in xgboost \n",
    "#with pseudo-huber loss \n",
    "kfold = KFold(n_splits=5, shuffle=True)\n",
    "alpha_value = [-4, -3, -2, -1, 0, 1, 2, 3, 4]\n",
    "maes = []\n",
    "\n",
    "for a in tqdm(alpha_value):\n",
    "    perform = []\n",
    "    for train_index, test_index in kfold.split(X):\n",
    "        train_x, valid_x = X[train_index], X[test_index]\n",
    "        train_y, valid_y = Y[train_index], Y[test_index]\n",
    "        model = xgb.XGBRegressor(objective=\"reg:pseudohubererror\", alpha = 10**a,reg_lambda = 0,seed=123)\n",
    "        model.fit(train_x, train_y)\n",
    "        pred_y = model.predict(valid_x)\n",
    "        perform.append(mean_absolute_error(valid_y, pred_y))\n",
    "    maes.append(np.mean(perform))    \n",
    "    \n",
    "\n",
    "print(maes)\n",
    "#\n",
    "#plot the graph of alpha_values and maes\n",
    "plt.plot(alpha_value, maes)\n",
    "plt.xlabel('alpha_value')\n",
    "plt.ylabel('Mean Absolute Error')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 9/9 [07:06<00:00, 47.38s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[1.8766753747075335, 1.767136296387664, 1.719292546723717, 1.706913355960673, 1.697004661283873, 1.7099746502725428, 1.723764870642946, 1.8054739861362101, 1.9458992653370153]\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "#use cross validation to test the optimal tree depth in xgboost with absolute error as the loss function\n",
    "kfold = KFold(n_splits=5, shuffle=True)\n",
    "tree_depths = [1,2,3,4,5,7,10,15,20]\n",
    "maes = []\n",
    "\n",
    "for h in tqdm(tree_depths):\n",
    "    perform = []\n",
    "    for train_index, test_index in kfold.split(X):\n",
    "        train_x, valid_x = X[train_index], X[test_index]\n",
    "        train_y, valid_y = Y[train_index], Y[test_index]\n",
    "        model = xgb.XGBRegressor(objective=\"reg:absoluteerror\", alpha = 1, max_depth = h,reg_lambda = 0,seed=123)\n",
    "        model.fit(train_x, train_y)\n",
    "        pred_y = model.predict(valid_x)\n",
    "        perform.append(mean_absolute_error(valid_y, pred_y))\n",
    "    maes.append(np.mean(perform))    \n",
    "    \n",
    "\n",
    "print(maes)\n",
    "#[1.8776232081766735, 1.769646219505216, 1.7208844670817656, 1.7074859514163023, 1.7027271100726615, 1.7082303992902044, 1.7386322177845002, 1.8395040007977055, 2.030162320631444]\n",
    "#the optimal tree depth is 5\n",
    "\n",
    "#plot the graph of tree_depths and maes\n",
    "plt.plot(tree_depths, maes)\n",
    "plt.xlabel('tree_depths')\n",
    "plt.ylabel('Mean Absolute Error')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 9/9 [01:50<00:00, 12.31s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[2.1132150486866683, 2.004721902186744, 1.9819564592650831, 1.9876500303827196, 1.9856478186735387, 1.9975593681382662, 2.041722501271953, 2.041820825402932, 2.0558383837576586]\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "#use cross validation to test the optimal tree depth in xgboost with pseudo-huber loss as the loss function\n",
    "kfold = KFold(n_splits=5, shuffle=True)\n",
    "tree_depths = [1,2,3,4,5,7,10,15,20]\n",
    "maes = []\n",
    "\n",
    "for h in tqdm(tree_depths):\n",
    "    perform = []\n",
    "    for train_index, test_index in kfold.split(X):\n",
    "        train_x, valid_x = X[train_index], X[test_index]\n",
    "        train_y, valid_y = Y[train_index], Y[test_index]\n",
    "        model = xgb.XGBRegressor(objective=\"reg:pseudohubererror\", alpha = 1000, max_depth = h,reg_lambda = 0,seed=123)\n",
    "        model.fit(train_x, train_y)\n",
    "        pred_y = model.predict(valid_x)\n",
    "        perform.append(mean_absolute_error(valid_y, pred_y))\n",
    "    maes.append(np.mean(perform))    \n",
    "    \n",
    "\n",
    "print(maes)\n",
    "#the optimal tree depth is 5\n",
    "\n",
    "#plot the graph of tree_depths and maes\n",
    "plt.plot(tree_depths, maes)\n",
    "plt.xlabel('tree_depths')\n",
    "plt.ylabel('Mean Absolute Error')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1.4332014997904543\n",
      "2.6194916280745137\n"
     ]
    }
   ],
   "source": [
    "part_answer = pd.read_csv(\"test_partial_answer.csv\", index_col='id')\n",
    "\n",
    "#retrain the XGboost model with full data\n",
    "model = xgb.XGBRegressor(objective=\"reg:absoluteerror\", alpha = 1,max_depth = 5,reg_lambda = 0,seed=123)\n",
    "model.fit(X, Y)\n",
    "print(np.mean(np.abs(model.predict(X) - Y)))\n",
    "\n",
    "df_test['Danceability'] = model.predict(X_test)\n",
    "df_test['Danceability'].to_csv('submission.csv', index=True)\n",
    "#print(df_test['Danceability'])\n",
    "#print(part_answer)\n",
    "\n",
    "pred_y_test = df_test[\"Danceability\"][range(18871,23485)]\n",
    "print(np.mean(np.abs(pred_y_test - part_answer[\"Danceability\"])))\n",
    "#8 2.55 2.56\n",
    "\n",
    "#MAE\n",
    "#1.4332014997904543\n",
    "#2.6194916280745137\n",
    "#public=2.2895\n",
    "#PHE\n",
    "#1.9597802909259263\n",
    "#2.7944495468245445\n",
    "#public=2.561\n",
    "\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 9/9 [2:09:27<00:00, 863.05s/it]   "
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[2.236924868957484, 2.1057658707047175, 2.034944670937682, 1.9354688410017473, 1.9039895165987186, 1.9004076878276064, 1.8959522422830521, 1.9036691904484566, 1.9183168316831682]\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "#use cross-validation to test the optimal tree_depth of the decision tree\n",
    "kfold = KFold(n_splits=5, shuffle=True)\n",
    "tree_depth = [1,2,3,5,6,7,8,9,10]\n",
    "maes = []\n",
    "\n",
    "for d in tqdm(tree_depth):\n",
    "    perform = []\n",
    "    for train_index, test_index in kfold.split(X):\n",
    "        train_x, valid_x = X[train_index], X[test_index]\n",
    "        train_y, valid_y = Y[train_index], Y[test_index]\n",
    "        model = DecisionTreeRegressor(max_depth=d, criterion='absolute_error', random_state=123)\n",
    "        model.fit(train_x, train_y)\n",
    "        pred_y = model.predict(valid_x)\n",
    "        perform.append(mean_absolute_error(valid_y, pred_y))\n",
    "    maes.append(np.mean(perform))    \n",
    "    \n",
    "\n",
    "print(maes)\n",
    "#[2.3200349446709376, 2.3150844496214327, 2.3071248301300717, 2.3063482818870122, 2.3026790914385558, 2.2291788002329644, 2.211852067559697, 2.2059794214715587, 2.1945253348864293, 2.1974956319161323, 2.111531741409435, 2.1046884100174723, 2.1061929722384, 2.100393127548049, 2.0960978450786256, 1.9522422830518347, 1.9285818287711125, 1.9474373907979032, 1.9511866627839254, 1.9626383226557949, 1.9196272568433315, 1.8867938264414676, 1.8980295088332362, 1.9062682003494467, 1.9005824111822947, 1.9111822947000583, 1.922029702970297, 1.9131236653077073, 1.911327897495632, 1.911677344205009]\n",
    "\n",
    "#plot the graph of tree_depth and cv_scores\n",
    "plt.plot(tree_depth, maes)\n",
    "plt.xlabel('Tree Depth')\n",
    "plt.ylabel('Mean Absolute Error')\n",
    "plt.show()\n",
    "#we can see that the optimal tree_depth is 5, so we use 5 as the tree_depth of the decision tree in the following part"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 8/8 [1:31:58<00:00, 689.81s/it]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[1.9233546884100174, 1.8669044845661038, 1.8172451951077462, 1.816421083284799, 1.8109007959619494, 1.8100446515239756, 1.8048888279831445, 1.8095821199767037]\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "#use cross-validation to test the optimal number of trees of the random forest\n",
    "kfold = KFold(n_splits=5, shuffle=True)\n",
    "tree_num = [1,2,5,10,12,15,17,20]\n",
    "maes = []\n",
    "\n",
    "for n in tqdm(tree_num):\n",
    "    perform = []\n",
    "    for train_index, test_index in kfold.split(X):\n",
    "        train_x, valid_x = X[train_index], X[test_index]\n",
    "        train_y, valid_y = Y[train_index], Y[test_index]\n",
    "        model = RandomForestRegressor(n_estimators=n,max_depth=7,criterion='absolute_error',n_jobs=-1,random_state=123)\n",
    "        model.fit(train_x, train_y)\n",
    "        pred_y = model.predict(valid_x)\n",
    "        perform.append(mean_absolute_error(valid_y, pred_y))\n",
    "    maes.append(np.mean(perform))    \n",
    "    \n",
    "\n",
    "\n",
    "print(maes)\n",
    "#[2.3200349446709376, 2.3150844496214327, 2.3071248301300717, 2.3063482818870122, 2.3026790914385558, 2.2291788002329644, 2.211852067559697, 2.2059794214715587, 2.1945253348864293, 2.1974956319161323, 2.111531741409435, 2.1046884100174723, 2.1061929722384, 2.100393127548049, 2.0960978450786256, 1.9522422830518347, 1.9285818287711125, 1.9474373907979032, 1.9511866627839254, 1.9626383226557949, 1.9196272568433315, 1.8867938264414676, 1.8980295088332362, 1.9062682003494467, 1.9005824111822947, 1.9111822947000583, 1.922029702970297, 1.9131236653077073, 1.911327897495632, 1.911677344205009]\n",
    "\n",
    "#plot the graph of tree_depth and cv_scores\n",
    "plt.plot(tree_num, maes)\n",
    "plt.xlabel('Tree Numbers')\n",
    "plt.ylabel('Mean Absolute Error')\n",
    "plt.show()\n",
    "#we can see that the optimal tree_\n",
    "\n",
    "# we can see when number of the decision tree is 15 ,the perform is good enough\n",
    "# so we use 15 as the number of the decision tree in the random forest model in the following part"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "  0%|          | 0/5 [00:00<?, ?it/s]"
     ]
    }
   ],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1.6933061541448262\n",
      "2.5123085050184897\n"
     ]
    }
   ],
   "source": [
    "\n",
    "#retrain the Random Forest model with full data\n",
    "model = RandomForestRegressor(n_estimators=15, max_depth=7 ,criterion='absolute_error',n_jobs=-1,random_state=123)\n",
    "model.fit(X, Y)\n",
    "part_answer = pd.read_csv(\"test_partial_answer.csv\", index_col='id')\n",
    "\n",
    "print(np.mean(np.abs(model.predict(X) - Y)))\n",
    "\n",
    "df_test['Danceability'] = model.predict(X_test)\n",
    "df_test['Danceability'].to_csv('submission.csv', index=True)\n",
    "#print(df_test['Danceability'])\n",
    "#print(part_answer)\n",
    "\n",
    "pred_y_test = df_test[\"Danceability\"][range(18871,23485)]\n",
    "print(np.mean(np.abs(pred_y_test - part_answer[\"Danceability\"])))\n",
    "#8 2.55 2.56\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "#run linear regression on training set\n",
    "model = LinearRegression()\n",
    "mae_score = []\n",
    "for i in tqdm(range(100)):\n",
    "    train_x, val_x, train_y, val_y = train_test_split(X, Y, test_size = 0.2)\n",
    "    model.fit(train_x, train_y)\n",
    "\n",
    "    #predict on validation set\n",
    "    pred_y = model.predict(val_x) \n",
    "    mae = np.mean(np.abs(pred_y - val_y))\n",
    "    mae_score.append(mae)\n",
    "print(np.mean(mae_score))\n",
    "#1.864922953017888\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "  0%|          | 0/20 [00:00<?, ?it/s]"
     ]
    }
   ],
   "source": [
    "#run Lasso regression on training set\n",
    "alphas = [0.001, 0.01, 0.1, 1, 10, 100, 1000]\n",
    "result = []\n",
    "for a in alphas:\n",
    "    model = Lasso(alpha=a,random_state=123)\n",
    "    mae_score = []\n",
    "    for i in tqdm(range(20)):\n",
    "        train_x, val_x, train_y, val_y = train_test_split(X, Y, test_size = 0.2)\n",
    "        model.fit(train_x, train_y)\n",
    "\n",
    "        #predict on validation set\n",
    "        pred_y = model.predict(val_x) \n",
    "        mae = np.mean(np.abs(pred_y - val_y))\n",
    "        mae_score.append(mae)\n",
    "    result.append(np.mean(mae_score))\n",
    "print(result)\n",
    "#[1.859802672727658, 1.8705790319054671, 2.026651088701079, 2.4506405558403506, 2.487390611341788, 2.4981356969718345, 2.488871803904156]\n",
    "#min = 1.859802672727658\n",
    "#0.001 has the best performance\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1.8440895291318975\n",
      "2.3580867874397318\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/opt/homebrew/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:631: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 1.028e+03, tolerance: 1.403e+01\n",
      "  model = cd_fast.enet_coordinate_descent(\n"
     ]
    }
   ],
   "source": [
    "#retrain the Lasso model with full data\n",
    "model = Lasso(alpha=0.01,random_state=123)\n",
    "model.fit(X, Y)\n",
    "part_answer = pd.read_csv(\"test_partial_answer.csv\", index_col='id')\n",
    "\n",
    "print(np.mean(np.abs(model.predict(X) - Y)))\n",
    "\n",
    "df_test['Danceability'] = model.predict(X_test)\n",
    "df_test['Danceability'].to_csv('submission.csv', index=True)\n",
    "#print(df_test['Danceability'])\n",
    "#print(part_answer)\n",
    "\n",
    "pred_y_test = df_test[\"Danceability\"][range(18871,23485)]\n",
    "print(np.mean(np.abs(pred_y_test - part_answer[\"Danceability\"])))\n",
    "#8 2.55 2.56"
   ]
  },
  {
   "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": "3.11.3"
  },
  "orig_nbformat": 4
 },
 "nbformat": 4,
 "nbformat_minor": 2
}