BAYESketball live win probability

The goal of this project is to create a live 'win probability' for the home team of any college basketball team during the 2021-22 season, in essence trying to reverse-engineer the "SUPER AWS ADVANCED ANALYTICS OMG" you see on TV.

Requirements for basic functionality

numpy, sqlite3, pandas and matplotlib.

Requirements for re-gathering the data and re-training the models:

pymc, formulae, bambi, arviz, cloudpickle, openpyxl, waybackpy

Running a simulation:

1. Extract 'data.7z' to the project root directory

2. Pick any game that ended in regulation from the 2021-22 NCAA Basketball season (One is pre-loaded), input it into line 11 of bayesketball-simulate.py.

3. Run 'bayesketball-simulate.py' with no arguments

4. The remaining scripts do not need to be re-run, they are for training the model and arranging the Data. most of them take forever and might nuke the databases if everything isn't prepared quite right.