nvsclub/SoccerActionsSim

SoccerActionEnv

A data-driven soccer environment built using event data. Compatible with OpenGYM API.

Instalation

- Requires Python 3.7 due to Tensorflow 1 requirements on stable_baselines3
- Run pip install -r requirements.txt
- Done!

Notebooks 00X - Procedures to build the underlying models

- 001 - Use this if you want to change the data used to build the simulator.
- 002 - Use this to change the way we build the shot model
- 003 - Use this to change the way we build the pass model

Notebooks 01X - Reinforcement learning

- 010 - Example of how to run a reinforcement learning on the environment
- 011 - Systematic testing of RL algorithms