A Modular Music Recommendation System using Unsupervised Learning and Spotify API Integration

This is a Python-based music suggestion system that utilizes machine learning algorithms combined with the Spotify API for aiding in song selection. It takes data inputs based on a list of songs, including name and year, from the users, and then combines it with Spotify's given metadata or audio feature. It makes use of K-Means clustering algorithm to group songs according to their numerical attributes like danceability, energy, and tempo, scaled with standard scaler. Dimensionality reduction techniques such as Principal Component Analysis (PCA) and t-Distributed Stochastic Neighbor Embedding (t-SNE) plot clusters in 2D. When a user requests a recommendation, the system computes a mean vector from the list of tracks given, computes the cosine distances to other tracks, and returns the closest tracks. The whole system is modular and can deal with the visualizations of clusters and edge cases like where a song does not exist in the dataset or library at Spotify. This integrates unsupervised learning, feature engineering, and interaction with real-time APIs to deliver a robust experience in recommending music.