Boosting_for_predictive_sufficiency

After installing basic packages such as numpy, scikit learn, xgboost, catboost, etc. Follow the steps below to reproduce the results in the paper.

1. Run CatBoost_Cluster_wrt_U.py to get results corresponding to Figure 2, Figure C1, Figure C2

2. Run XGBoost_Cluster_wrt_U.py to get results corresponding to Figure C3, Figure C4

3. Run XGBoost_Cluster_vs_MSE.py to get results corresponding to Figure 3.

4. Run california_housing.py and newsgroups.py to get results corresponding to Figure 4 and Figure 5

5. Run california_housing_comparison.py and newsgroups_comparison.py to get results corresponding to Table 1

If you use this project, please cite:

@inproceedings{
reddy2026boosting,
title={Boosting for Predictive Sufficiency},
author={Abbavaram Gowtham Reddy and Rajeev Verma and Celia Rubio-Madrigal and Krikamol Muandet and Rebekka Burkholz},
booktitle={The Fourteenth International Conference on Learning Representations},
year={2026},
url={https://openreview.net/forum?id=1mQT8PXIy8}
}