PeMS Regression

pems-regression is a benchmark suite for graph node regression with uncertainty.
It was originally derived from the California PeMS (Performance Measurement System) in Borovitskiy et al. (2021).
This project provides processed data, evaluation metrics, plotting tools, and baseline models.

Installation

Install via:

pip install git+https://github.com/vabor112/pems-regression.git

Usage

See the example notebooks in the notebooks folder.

Contributing

If you would like to contribute to the project, please ensure that make lint reports no issues before submitting a pull request.

Citation

If you use this benchmark in your research, please cite:

@misc{borovitskiy2025,
    title={PeMS Regression: A Benchmark Suite for Node Regression with Uncertainty},
    author={Viacheslav Borovitskiy},
    year={2025},
    url={https://github.com/vabor112/pems-regression}
}