repo for the source code of OASiS (Off-the-ball Action Significance Score), presented in SSAC 2020 Hackathon.
Full presentation available at https://youtu.be/52hivyFDmhs?t=2000
qingbowang_ssac_hackathon_presentation.pptx: describes the idea of OASiS
( qingbowang_ssac_hackathon_presentation.pdf: is the pdf version )
signif_action_example.mov: is the full version of the movie (tracking the players) presented in the conference
demo_data_generation.ipynb: generates the data used to perform the OASiS analysis
demo_data_exploration.ipynb: generates most of the figures used in the presentation
demo_movie_visualization.ipynb: generates the movie used in the presentation
demo_aggregate_oasis_allgames.ipynb: generates the OASiS across all the >10 different games provided in the datathon (Work in progress...)
The Datathon took place in part of MIT Sloan Sports Analytics Conference 2020 (SSAC2020) and was supported by ESPN and Shottracker.
All the analysis was performed in google colab environment, and the data was provided by Shottracker repo.
For setting up the environment, please see the google colab documentation as well as the Shottracker repo.
The demo shown here are prototypes, and more sophisticated, scalable pipeline is under development. Suggestions / comments are welcome!
Contact: Twitter @qbw_128 email: qingbo_wang [at] g.harvard.edu