A Spotify tracks analysis and success prediction for my CS109a final project at Harvard University.
Data is from Kaggle obtained using Spotify API.
Each track's features are explained on Spotify's Web API Reference.
This project involves binary and multi-class prediction for each song's popularity. Methods used are logistic regression, decision tree, k-nearest neighbors, random forest, and neural network.
Full report of our methods and results can be found here.
