This is my team's final machine learning project; our team was comprised of myself, Rebecca Chang, Jung-Yoon (Claire) Choi, and Brett Rado.
The original dataset, before using Spotipy to get additional predictors, can be found here: https://www.kaggle.com/theoverman/the-spotify-hit-predictor-dataset.
You will find folders with the Jupyter notebooks used to clean the data, create descriptive analytics/visualizations, and create models along with our CSV files and the script used to turn the original Kaggle dataset into the one we used for the project.
The models notebooks are broken up into pre-tuning, tuning hyperparameters using GridSearch, and tuning hyperparameters using RandomSearch.
If you have suggestions or want to build on this or use it for one of your projects, just be sure to cite this repository and the original Kaggle dataset.
If you want to check out our presentation that highlights the challenges that came with the project, performance, and other insights, check out the attached .pdf or .pptx files.