nba-prediction-models 🏀

about

This repository contains historic NBA game data and PyTorch-based machine learning models to make predictions on upcoming NBA games, using the novel approach of modelling the recent form of the teams playing as a sequential dataset of game statistics.

structure

The projects core modules and interfaces are as follows:

• train.ipynb notebook: Code used to train the models. The notebook has full instructions, and can be used to perform your own training on the implmented model architectures.
• predict.ipynb notebook: Interface to allow new result prediction from the pre-trained models. The notebook gives full instructions, and can be used to define your own dataset for a series of NBA game results, and then make a prediction on an upcoming game.
• data/* module: This is where the sequential datasets for training and evaluation are constructed, from a publicly available database of NBA game statistics. Checkout it's readme for more information.
• src/* module: This contains the underlying code for training and evaluating the models.
• output/* module: Contains any output produced from training, including various pretrained models.

installation

Clone the repository code:

git clone https://github.com/itsluketwist/nba-prediction-models.git

Once cloned, install the requirements locally in a virtual environment:

python -m venv .venv

. .venv/bin/activate

pip install -r requirements.lock

development

Install and use pre-commit to ensure code is in a good state:

pre-commit install

pre-commit autoupdate

pre-commit run --all-files

Use uv for dependency management, first add to requirements.txt. Then install uv and version lock with:

pip install uv

uv pip compile requirements.txt -o requirements.lock