Unsupervised_Machine_Learning

Project Overview

In this project, I clustered 5,000 Spotify songs using unsupervised machine learning.

Methodology

1. Data Cleaning: Cleaned the data to remove duplicates and empty values.
2. Feature Selection: Chose features like danceability, valence, loudness, energy, acousticness, etc., for clustering.
3. Clustering: Used KMeans for clustering the songs based on the selected features.
4. Dimensionality Reduction: Applied PCA to enhance clustering efficiency and interpretability.
5. Data Standardization: Standardized the data using MinMaxScaler to scale features between 0 and 1.
6. Determining Clusters: Used the Elbow Method based on Inertia score and Silhouette score to determine the optimal number of clusters.
7. Playlist Creation: Generated playlists based on KMeans clustering of similar songs closest to their cluster centers.