Initially inspired by this kaggle project. This repo hosts a few things:
- Collection of NBA game data
- Exploration of said nba game data
- Model generation to predict winners of nba games
- Analysis of model performance with betting odds data to predict performance
- A web application that displays game winners for the current day (COMING SOON)
Using uv to manage dependencies.
uv sync
uv run jupyter server
This will kick off the jupyter notebook server to run and explore the data notebooks used to generate nba game winner predictions.
Advised that you start in notebooks/00_init.ipynb
scrapy is used to retrieve the team, game, and betting odds data.
Game data is collected from the nba stats website.
Odds data is collected from sportsbookreview
To collect the data you can run the 00_init.ipynb notebook, or run scrapy commands for the various scrapers (NOTE: These can take a long time!)
ex: uv run scrapy crawl nba_teams
Model exploration happens in the notebook dir. Things like feature analysis, generation, model creation, profit calculation, etc is in these notebooks.
To run the webapp locally, run uv run uvicorn --reload web.app:main. It will be available at localhost:8000
Check out the issues and milestones