Song Recommendation System Using K-Means Clustering and Spotify API

Overview

This project builds a song recommendation system by clustering a collection of songs based on their audio features. The dataset was created by fetching song data using the Spotify API and clustering was performed using K-Means. The system enables more accurate song recommendations by limiting recommendations to songs within the same cluster, which share similar audio characteristics.

Key Features

• API Integration: Used Spotify API to retrieve audio features like danceability, energy, loudness, and more for a large number of songs.
• Unsupervised Learning: Applied K-Means Clustering to group songs into clusters based on their audio attributes.
• Feature Evaluation: Conducted ANOVA to determine which audio features significantly vary across clusters.
• Optimal Clustering: Determined the best number of clusters using the Elbow Method and Silhouette Score for better recommendation accuracy.

Technologies Used

• Programming Language: Python
• Libraries: pandas, NumPy, seaborn, matplotlib, scikit-learn, statsmodels
• API: Spotify API
• Machine Learning: KMeans Clustering, ANOVA

Dataset

• Source: Data was collected using the Spotify API. The dataset includes a wide range of audio features for each song.
• Key Features: Danceability Energy Loudness Tempo Acousticness Instrumentalness Speechiness Popularity Length

Spotify API Setup:

You will need to create a Spotify Developer account to obtain your API key and client secret. Create a config.py file and add your credentials:

python
    client_id = 'your-client-id'
    client_secret = 'your-client-secret'

Results

• Optimal Clustering: The optimal number of clusters was determined using the Elbow Method and Silhouette Scores.
• Cluster Analysis: Audio features like length and time signature showed significant differences between clusters, as confirmed by ANOVA tests.

Future Work

• Improve the recommendation algorithm by adding more audio features and refining the clustering process.
• Integrate a web interface using Flask to allow users to input a song and receive recommendations in real-time.