gab992/Content-Based-Live-Music-Recommender
A playlist recommender that recommends songs by artists with upcoming shows in a user's city and bases those recommendations on audio features. Recommendations use a combination of pre-defined Spotify audio features and audio features (MFCCs) extracted from mp3s using the LibRosa Python library. Genre clusters are created using PCA dimensionality reduction and K-Means clustering on MFCCs. Data sources include the Spotify and Songkick APIs as well as web-scraped information from Pitchfork and YouTube mp3s.
Jupyter Notebook