Markov Decision Processes with Dynamic Transition Probabilities: An Analysis of Shooting Strategies in Basketball

A Companion Repository

This repository contains a one game sample of optical player tracking data courtesy of STATS LLC from the 2012-2013 NBA regular season between the Miami Heat and the Brooklyn Nets, formatted according to the methods described in our paper. The purpose of this repository is to provide a reproducible, simplified analysis of the methods introduced in Markov Decision Processes with Dynamic Transition Probabilities: An Analysis of Shooting Strategies in Basketball.

The R markdown document walkthrough.Rmd shows how to simulate plays from this game using the models described in Section 3 of the paper for each team's MDP and the simulation algorithm defined in Section 4. The document concludes by testing an altered policy for the Miami Heat and comparing the simulations under this altered policy to the on-poicy simulations. A knit pdf of the walkthrough document has been uploaded as well.