Skip to content

A machine learning library for regression, which implements a new formulation of gradient boosting.

License

Notifications You must be signed in to change notification settings

a-wozniakowski/scikit-physlearn

Repository files navigation

Scikit-physlearn

SOTA Documentation Status PyPI

Documentation | Base boosting

Scikit-physlearn amalgamates Scikit-learn, LightGBM, XGBoost, CatBoost, and Mlxtend regressors into a flexible framework that:

  • Follows the Scikit-learn API.
  • Processes pandas data representations.
  • Solves single-target and multi-target regression tasks.
  • Interprets regressors with SHAP.

Additionally, the library contains the official implementation of base boosting, which is a reformulation of gradient boosting that

  • Regards predictions from any regression model as an inductive bias.

In contrast, gradient boosting regards the prediction from a constant model as an inductive bias.

  • Consequently, base boosting generalizes Tukey’s methods of twicing, thricing, and reroughing, as gradient boosting works with a variety of fitting criterion.

The machine learning library was started by Alex Wozniakowski during his graduate studies at Nanyang Technological University.

Installation

Scikit-physlearn can be installed from PyPI:

pip install scikit-physlearn

To build from source, follow the installation guide.

Citation

If you use this library, please consider adding the corresponding citation:

@article{wozniakowski_2021_boosting,
  title={A new formulation of gradient boosting},
  author={Wozniakowski, Alex and Thompson, Jayne and Gu, Mile and Binder, Felix C.},
  journal={Machine Learning: Science and Technology},
  volume={2},
  number={4},
  year={2021}
}

About

A machine learning library for regression, which implements a new formulation of gradient boosting.

Topics

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages