/multiply_robust

An implementation of the multiply robust estimation described in the paper "Multiply Robust Estimation for Local Distribution Shifts with Multiple Domains" (ICML 2024).

Primary LanguagePythonMIT LicenseMIT

Multiply Robust Estimation

This software package implements the multiply robust method proposed in the paper "Multiply Robust Estimation for Local Distribution Shifts with Multiple Domains".

@inproceedings{wilkins-reeves2024multiply,
    author = {Wilkins-Reeves, Steven and Chen, Xu and Ma, Qi and Agarwal, Christine and Hofleitner, Aude},
    title = {{Multiply Robust Estimation for Local Distribution Shifts with Multiple Domains}},
    booktitle = {Proceedings of the 41st International Conference on Machine Learning. (ICML 2024)},
    year = {2024},
}

Usage

  • estimators: implement the multiply robust estimation method
  • utils: implement utility functions including data preprocessing and segment clustering algorithms
  • weights: implement different importance weighting methods to adjust for covariate shifts and label shifts

A demo notebook will be added!

Installation

Installation Requirements

  • Python >= 3.6
  • numpy >= 1.24.4
  • sklearn >= 1.2.2
  • xgboost >= 1.2.0
  • scipy >= 1.10.1
  • cvxopt >= 1.2.7

Contribute

See the CONTRIBUTING file for how to help out.

License

multiply_robust package is MIT licensed, as found in the LICENSE file.