master.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 17,
   "id": "872d7e83",
   "metadata": {},
   "outputs": [],
   "source": [
    "# 1. import libraries\n",
    "from bs4 import BeautifulSoup\n",
    "import requests\n",
    "import pandas as pd\n",
    "import spotipy\n",
    "import json\n",
    "from spotipy.oauth2 import SpotifyClientCredentials\n",
    "import config\n",
    "from sklearn.preprocessing import StandardScaler\n",
    "from sklearn.cluster import KMeans\n",
    "from IPython.display import IFrame\n",
    "#Initialize SpotiPy with user credentias\n",
    "sp = spotipy.Spotify(auth_manager=SpotifyClientCredentials(client_id= config.client_id,\n",
    "                                                           client_secret= config.client_secret))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "id": "04187015",
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "def recommender(): \n",
    "     df = pd.read_csv(\"billboard100_data.csv\") # data of top 100 songs\n",
    "     df['title'] = df['title'].map(str.lower) # converting the entries to lower\n",
    "     df['author'] = df['author'].map(str.lower) # converting the entries to lower\n",
    "\n",
    "     x = input('Enter a song you like: ')\n",
    "     x_artist = df.loc[df['title'].map(str.lower)  == x.lower(), 'author'] # artist corresponding to song x from billboard100_data.csv\n",
    "\n",
    "     results = sp.search(q=x,limit=3) # gives the deatils of the inputed song x\n",
    "     track_id=results[\"tracks\"][\"items\"][0][\"id\"] # track_id of the song x\n",
    "     display(IFrame(src=\"https://open.spotify.com/embed/track/\"+track_id,\n",
    "               width=\"320\",\n",
    "               height=\"80\",\n",
    "               frameborder=\"0\",\n",
    "               allowtransparency=\"true\",\n",
    "               allow=\"encrypted-media\",))  \n",
    "\n",
    "     ### When the inputed song belong to the top100 songs, i.e., in df \n",
    "     if x.lower() in df.values:   # converts the entered x into all lower case. When x is in the billboard100\n",
    "        rec_song = df['title'].sample().to_string(index=False) # recommendation of song. Removed the indices as they are not required \n",
    "        rec_song_results = sp.search(q=rec_song,limit=3) # gives the deatils of rec_song\n",
    "        track_id_rec=rec_song_results[\"tracks\"][\"items\"][0][\"id\"] # track_id of the song rec_song \n",
    "        rec_artist = df.loc[df['title'] == rec_song, 'author'].to_string(index=False) # artist of the recommended song\n",
    "        print(\"You might like this song: \" + rec_song.title() + \" by \" + rec_artist.title())  # .title() makes the first alphabet in every word to upper   \n",
    "        display(IFrame(src=\"https://open.spotify.com/embed/track/\"+track_id_rec,\n",
    "               width=\"320\",\n",
    "               height=\"80\",\n",
    "               frameborder=\"0\",\n",
    "               allowtransparency=\"true\",\n",
    "               allow=\"encrypted-media\",)) \n",
    "     ### When the inputed song does not belong to top100, we give a suggestion from out database, i.e., df_spotify\n",
    "     else:   \n",
    "        df_spotify = pd.read_csv(\"spotify_data.csv\") # collected data of songs from spotify (unclustered)\n",
    "        X = df_spotify.drop(['id', 'duration_ms'], axis=1) # spotify data without id and duration (because X needs to be numerical)\n",
    "        scaler = StandardScaler() # for scaling of X\n",
    "        scaler_fitted = scaler.fit(X) # we need scaler_fitted for transforming/scaling the song which is inputed, i.e., x \n",
    "        X_scaled = scaler_fitted.transform(X)\n",
    "        X_scaled_df = pd.DataFrame(X_scaled, columns = X.columns) # scaled the numerical feature data X\n",
    "        kmeans = KMeans(n_clusters=4, random_state=1234) # number of clusters is 4\n",
    "        kmeans.fit(X_scaled_df.values) # for prediction of cluster of the inputed song (x), we need kmean fitted to X_scaled (our data base)\n",
    "    \n",
    "       ### from inputed song to extracting features and scaling\n",
    "        x_af = sp.audio_features(track_id) # extracting audio features, i.e., danceability, tempo etc\n",
    "        X_af = pd.DataFrame(x_af) # audio features (af) as a dataframe\n",
    "        X_af_num = X_af.drop(['key', 'mode', 'type', 'id', 'uri', 'track_href', \n",
    "                          'analysis_url', 'duration_ms', 'time_signature'], axis=1) # droping unwanted columns\n",
    "        X_af_num_scaled = scaler_fitted.transform(X_af_num) # scaled the inputed song to the parameters from scaler_fitted \n",
    "    \n",
    "       ### predicting the song cluster\n",
    "        song_cluster = kmeans.predict(X_af_num_scaled) # cluster of the inputed song based on kmeans fitted to our database X_scaled_df\n",
    "    \n",
    "       ### checking in which cluster the inputed song belongs\n",
    "        df_cluster = pd.read_csv(\"clustered_data.csv\") # data of clustered (4) songs from spotify \n",
    "    \n",
    "        n_clusters = 4  # four clusters, i.e., 0,1,2,3\n",
    "        for i in  range(0, n_clusters):\n",
    "            if song_cluster == i:  \n",
    "                mask = df_cluster['cluster'] == i\n",
    "                selected_rows = df_cluster[mask]\n",
    "                random_row = selected_rows.sample()\n",
    "                track_id_rand = random_row['id'].to_string(index=False)\n",
    "                track = sp.track(track_id_rand)\n",
    "                print('You might like this song: ' + track[\"name\"]+ ' by ' + track[\"album\"][\"artists\"][0][\"name\"]) \n",
    "                display(IFrame(src=\"https://open.spotify.com/embed/track/\"+track_id_rand,\n",
    "                   width=\"320\",\n",
    "                   height=\"80\",\n",
    "                   frameborder=\"0\",\n",
    "                   allowtransparency=\"true\",\n",
    "                   allow=\"encrypted-media\",))  \n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "id": "7a4c93dc",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Enter a song you like: flowers\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "\n",
       "        <iframe\n",
       "            width=\"320\"\n",
       "            height=\"80\"\n",
       "            src=\"https://open.spotify.com/embed/track/0yLdNVWF3Srea0uzk55zFn?frameborder=0&allowtransparency=true&allow=encrypted-media\"\n",
       "            frameborder=\"0\"\n",
       "            allowfullscreen\n",
       "            \n",
       "        ></iframe>\n",
       "        "
      ],
      "text/plain": [
       "<IPython.lib.display.IFrame at 0x2198ef53130>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "You might like this song: Tomorrow 2 by Glorilla & Cardi B\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "\n",
       "        <iframe\n",
       "            width=\"320\"\n",
       "            height=\"80\"\n",
       "            src=\"https://open.spotify.com/embed/track/7KXVIAuw3m2hxZanrpSXS3?frameborder=0&allowtransparency=true&allow=encrypted-media\"\n",
       "            frameborder=\"0\"\n",
       "            allowfullscreen\n",
       "            \n",
       "        ></iframe>\n",
       "        "
      ],
      "text/plain": [
       "<IPython.lib.display.IFrame at 0x2198ef53220>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "recommender()"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "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": "3.9.13"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}