This is the implementation of our paper: Quang-Duy Tran, Bao Duong, Phuoc Nguyen, and Thin Nguyen. Robust Estimation of Causal Heteroscedastic Noise Models (ROCHE). In Proceedings of the 2024 SIAM International Conference on Data Mining (SDM), 2024.
The configuration for the conda environment is available at conda.yml.
To run the experiments, use the run.py with the following configurations:
--method roche: using ROCHE as the method,--data [DATASET]: the benchmark dataset.
To see all available configurations for each experiment, run the python file with -h or --help.
- For baselines implemented in R (CAM, GRCI, IGCI, QCCD, and RESIT), the results are available in
baseline_results/. - For baselines implemented in Python (CGNN, HECI, and LOCI), the results are available in
baseline_results_py/. - For ROCHE, the results are available in
results/.
This code is based on the implementation of Location-Scale Causal Inference (LOCI).
If you find our code helpful, please cite us as:
@inproceedings{tran2024robust,
author = {Tran, Quang-Duy and Duong, Bao and Nguyen, Phuoc and Nguyen, Thin},
booktitle = {Proceedings of the 2024 SIAM International Conference on Data Mining (SDM)},
title = {Robust Estimation of Causal Heteroscedastic Noise Models},
year = {2024},
pages = {788--796},
}