shaunmckellar9 / spotify-recommender

Shaun's Spotify Recommender System

Project Overview

Welcome to my Spotify Recommender System! This project represents a blend of my love for music and my passion for data analysis. Growing up in a household where music was the heartbeat of our daily lives, I’ve always been curious about exploring new tracks and artists. This project aims to answer a simple yet intriguing question: "Who and what does Shaun listen to the most on Spotify?"

Table of Contents Introduction Project Goals Data Collection Data Analysis Recommendations Cluster Analysis Audio Features Analysis Playlist Creation Future Work Usage Instructions Contact

Introduction This project leverages the Spotify API to fetch my liked songs, analyze them, and provide personalized recommendations. It showcases my ability to connect to APIs, clean and analyze data, and visualize insights.

Project Goals Understand my musical preferences: Identify the genres, artists, and characteristics that dominate my playlist. Build a recommendation system: Suggest new tracks based on my listening history. Analyze audio features: Gain insights into the audio properties that appeal to me the most. Create a dynamic playlist: Automatically update my recommendations with a weekly cron job.

Data Collection Spotify API Using the Spotify API, I extracted data for over 3,700 liked songs, including:

Track names Artists Albums Popularity scores Audio features (e.g., danceability, energy, tempo, etc.) python

Copy code import spotipy from spotipy.oauth2 import SpotifyOAuth

Spotify authentication

sp = spotipy.Spotify(auth_manager=SpotifyOAuth( client_id='YOUR_CLIENT_ID', client_secret='YOUR_CLIENT_SECRET', redirect_uri='YOUR_REDIRECT_URI', scope='user-library-read user-top-read playlist-modify-public' ))

Fetch liked songs

results = sp.current_user_saved_tracks(limit=50) liked_songs = results['items']

Continue fetching more liked songs

while results['next']: results = sp.next(results) liked_songs.extend(results['items'])

Data Analysis Genre and Feature Analysis Analyzed the genres and audio features to understand my music preferences. Key insights include:

Dominant genres: Hip hop and rap were the most prevalent.

Audio characteristics: Tracks with high danceability and energy were preferred, indicating a penchant for lively and engaging music.

Python Copy code

Extract relevant features and metadata

song_data = [] for item in liked_songs: track = item['track'] track_info = { 'track_id': track['id'], 'track_name': track['name'], 'artist_name': track['artists'][0]['name'], 'popularity': track['popularity'], 'album': track['album']['name'] } song_data.append(track_info)

Convert to DataFrame

df_songs = pd.DataFrame(song_data) Recommendations Developed a recommendation engine using TF-IDF and cosine similarity to suggest new songs based on my current playlist. Learning about these techniques and implementing them was a rewarding challenge.

python Copy code from sklearn.feature_extraction.text import TfidfVectorizer from sklearn.metrics.pairwise import cosine_similarity

Combine metadata for each song

df_songs['metadata'] = df_songs[['artist_name', 'genres', 'popularity']].apply(lambda x: ' '.join(map(str, x)), axis=1)

Calculate TF-IDF and cosine similarity

tfidf_vectorizer = TfidfVectorizer(stop_words='english') tfidf_matrix = tfidf_vectorizer.fit_transform(df_songs['metadata']) cosine_sim = cosine_similarity(tfidf_matrix, tfidf_matrix)

Get recommendations

def get_recommendations(track_name, similarity_df, df_songs, num_recommendations=5): similar_tracks = similarity_df[track_name].sort_values(ascending=False)[1:num_recommendations+1] similar_track_names = similar_tracks.index return df_songs[df_songs['track_name'].isin(similar_track_names)] Cluster Analysis Using PCA and KMeans clustering, I grouped my songs into distinct clusters. Each cluster represented unique musical themes, helping me understand the diversity in my playlist.

python Copy code from sklearn.decomposition import PCA from sklearn.cluster import KMeans

Perform PCA and KMeans clustering

pca = PCA(n_components=2) principal_components = pca.fit_transform(df_songs[['danceability', 'energy', 'tempo', 'valence', 'acousticness']]) df_songs['pca1'] = principal_components[:, 0] df_songs['pca2'] = principal_components[:, 1]

kmeans = KMeans(n_clusters=5, random_state=42) df_songs['cluster'] = kmeans.fit_predict(df_songs[['danceability', 'energy', 'tempo', 'valence', 'acousticness']]) Audio Features Analysis Conducted a detailed analysis of audio features across my liked tracks and recommended tracks. This analysis provided insights into the characteristics that define my musical taste.

python Copy code

Fetch audio features for tracks

def fetch_audio_features(sp, track_ids): audio_features = [] for i in range(0, len(track_ids), 50): batch = track_ids[i:i+50] audio_features.extend(sp.audio_features(batch)) return audio_features

track_ids = df_songs['track_id'].tolist() audio_features = fetch_audio_features(sp, track_ids)

Convert audio features to DataFrame

df_audio_features = pd.DataFrame(audio_features) Playlist Creation Created a new playlist with the top recommendations and set up a cron job to update it weekly based on my latest liked tracks. This dynamic playlist ensures I always have fresh music to explore.

python Copy code

Create a new playlist

user_id = sp.current_user()['id'] playlist_name = 'Shaun New Recommended Playlist' new_playlist = sp.user_playlist_create(user_id, playlist_name, public=True)

Add recommended tracks to the playlist

track_uris = refined_recommendations['track_id'].apply(lambda x: f'spotify:track:{x}').tolist() sp.playlist_add_items(new_playlist['id'], track_uris) Future Work Weekly Updates: Implement a cron job to update my playlist with new recommendations based on my recent likes. Improved Recommendations: Continuously refine the recommendation algorithm to incorporate more audio features and improve accuracy. Usage Instructions Clone the repository: bash Copy code git clone https://github.com/shaunmckellarjr/Spotify-Recommender-System.git Install the required libraries: bash Copy code pip install -r requirements.txt Run the script: bash Copy code python spotify_recommender.py Contact Feel free to reach out if you have any questions or suggestions!

Email: [email protected] LinkedIn: Shaun McKellar Jr