CS4641 Machine Learning Project

This is a collaboration between Nathan Luskey, Reagan Matthews, Dylan Reese, & David Wen.

Project Overview

Our goal is to match your music tastes for customized running playlists. For the deliverables see our final report & presentation

Problem & Background

Who?

We want to help you find songs to match your workout tempo and personal taste of music.

How?

We used 2 different methods:

• Unsupervised: K-means clustering + PCA Optimized via the elbow method shown below: All code can be seen in the unsupervisedLearning directory.

• Supervised: Decision Tree with $\alpha$ Pruning and Genre as Label All code can be seen in the supervisedLearning directory.

Data

• The million songs data. All code can be seen in the SQLTesting directory.

Packages

• The data used in this project is obtained with dolthub, so you will need to install dolt and doltpy.
• The machine learning was done using scikit
• Visualizations used pyplot
• Created csv using pandas