ShotDeterminantsTrain

Poster: http://www.stat.cmu.edu/cmsac/poster2020/posters/Varadarajan-NBAShotProb.pdf

If you would like to see how the data is processed please visit the https://github.com/avyayv/ShotDeterminantsData.

You can simply run the Model.ipynb notebook, which will handle the training of the XGBoost model directly from the output of the DataProcessing notebook. The output of the DataProcessing notebook is a part of the repository already, so it is unnescassary to run it again, unless you would simply like to reproduce the results.

File Contents

1. trainXAll.csv Extracted information for every shot from PBP and Tracking Data (including defender distance, shot distance, player id, and defender id.
2. trainYAll.csv For the trainX, was the shot made or missed
3. bio_data.csv Includes height data for the 2015-16 season, which is an input into the model (defender height - shooter height is the input)
4. summary_data.csv Includes 3PT% for the players in the 2015-16 season.

All of the above files are generated using this repository https://github.com/avyayv/ShotDeterminantsData

5. Model.ipynb The actual model + interpretation of the model.