Scalable Counterfactual Distribution Estimation in Multivariate Causal Models

Thong Pham, Shohei Shimizu, Hideitsu Hino, Tam Le
Proceedings of the Third Conference on Causal Learning and Reasoning, PMLR 236:1118-1140, 2024.

Abstract

We consider the problem of estimating the counterfactual joint distribution of multiple quantities of interests (e.g., outcomes) in a multivariate causal model extended from the classical difference-in-difference design. Existing methods for this task either ignore the correlation structures among dimensions of the multivariate outcome by considering univariate causal models on each dimension separately and hence produce incorrect counterfactual distributions, or poorly scale even for moderate-size datasets when directly dealing with such a multivariate causal model. We propose a method that alleviates both issues simultaneously by leveraging a robust latent one-dimensional subspace of the original high-dimension space and exploiting the efficient estimation from the univariate causal model on such space. Since the construction of the one-dimensional subspace uses information from all the dimensions, our method can capture the correlation structures and produce good estimates of the counterfactual distribution. We demonstrate the advantages of our approach over existing methods on both synthetic and real-world data.

Cite this Paper

BibTeX
@InProceedings{pmlr-v236-pham24a, title = {Scalable Counterfactual Distribution Estimation in Multivariate Causal Models}, author = {Pham, Thong and Shimizu, Shohei and Hino, Hideitsu and Le, Tam}, booktitle = {Proceedings of the Third Conference on Causal Learning and Reasoning}, pages = {1118--1140}, year = {2024}, editor = {Locatello, Francesco and Didelez, Vanessa}, volume = {236}, series = {Proceedings of Machine Learning Research}, month = {01--03 Apr}, publisher = {PMLR}, pdf = {https://proceedings.mlr.press/v236/pham24a/pham24a.pdf}, url = {https://proceedings.mlr.press/v236/pham24a.html}, abstract = {We consider the problem of estimating the counterfactual joint distribution of multiple quantities of interests (e.g., outcomes) in a multivariate causal model extended from the classical difference-in-difference design. Existing methods for this task either ignore the correlation structures among dimensions of the multivariate outcome by considering univariate causal models on each dimension separately and hence produce incorrect counterfactual distributions, or poorly scale even for moderate-size datasets when directly dealing with such a multivariate causal model. We propose a method that alleviates both issues simultaneously by leveraging a robust latent one-dimensional subspace of the original high-dimension space and exploiting the efficient estimation from the univariate causal model on such space. Since the construction of the one-dimensional subspace uses information from all the dimensions, our method can capture the correlation structures and produce good estimates of the counterfactual distribution. We demonstrate the advantages of our approach over existing methods on both synthetic and real-world data.} }
Endnote
%0 Conference Paper %T Scalable Counterfactual Distribution Estimation in Multivariate Causal Models %A Thong Pham %A Shohei Shimizu %A Hideitsu Hino %A Tam Le %B Proceedings of the Third Conference on Causal Learning and Reasoning %C Proceedings of Machine Learning Research %D 2024 %E Francesco Locatello %E Vanessa Didelez %F pmlr-v236-pham24a %I PMLR %P 1118--1140 %U https://proceedings.mlr.press/v236/pham24a.html %V 236 %X We consider the problem of estimating the counterfactual joint distribution of multiple quantities of interests (e.g., outcomes) in a multivariate causal model extended from the classical difference-in-difference design. Existing methods for this task either ignore the correlation structures among dimensions of the multivariate outcome by considering univariate causal models on each dimension separately and hence produce incorrect counterfactual distributions, or poorly scale even for moderate-size datasets when directly dealing with such a multivariate causal model. We propose a method that alleviates both issues simultaneously by leveraging a robust latent one-dimensional subspace of the original high-dimension space and exploiting the efficient estimation from the univariate causal model on such space. Since the construction of the one-dimensional subspace uses information from all the dimensions, our method can capture the correlation structures and produce good estimates of the counterfactual distribution. We demonstrate the advantages of our approach over existing methods on both synthetic and real-world data.
APA
Pham, T., Shimizu, S., Hino, H. & Le, T.. (2024). Scalable Counterfactual Distribution Estimation in Multivariate Causal Models. Proceedings of the Third Conference on Causal Learning and Reasoning, in Proceedings of Machine Learning Research 236:1118-1140 Available from https://proceedings.mlr.press/v236/pham24a.html.

Related Material