A supervised machine learning project that attempts to build a model capable of predicting a song's popularity using audio features provided by the Spotify database.
For the full project, check out the Jupyter notebook titled Predicting_Song_Popularity.ipynb. For a more high-level overview, check out the pdf files.
Additionally, the reports folder contains notebooks with further details and analysis for the various stages of the project.
This project uses different approaches, including linear regression and classification, to try and predict which songs are popular. Using ensemble methods, a final accuracy of 86% was achieved and a recall of 0.76 for predicting songs of the popular class.