daorichard/spotify_machine_learning

Final Data Science Project - Spotify Feature Machine Learning Model

DS3K

This README attempts to explain the structure of our DS3000 project.

data

• Contains both raw and processed data used in our project.
• The subdirectories with group member names are their individual data records from Spotify.
• The subdirectory group includes the final processed data used to train our models.
  • You can see most of its creation in the transformation directory.

eda

• Contains initial exploration of data, as well as some early steps of data conglomeration, transformation, and visualization
• Much of the data included in these files was not used in final model training, as we originally featurized our entire streaming history

transformation

• Contains code that was used to prepare the master table of our streaming libraries and create the train/test split.
• Data is transformed (encoded / scaled) in these notebooks, then saved to data/

classification_models

• Contains notebooks where each of our model implementations were created
• Additionally includes model comparison visualizations.

Upper level files

• api-setup.py - used to get API keys from a file ignored by git
• requirements.txt - some of the libraries used in our analysis
  • Probably isn't up to date!