Transformer-Based Spatial-Temporal Counterfactual Outcomes Estimation

He Li, Haoang Chi, Mingyu Liu, Wanrong Huang, Liyang Xu, Wenjing Yang
Proceedings of the 42nd International Conference on Machine Learning, PMLR 267:34185-34208, 2025.

Abstract

The real world naturally has dimensions of time and space. Therefore, estimating the counterfactual outcomes with spatial-temporal attributes is a crucial problem. However, previous methods are based on classical statistical models, which still have limitations in performance and generalization. This paper proposes a novel framework for estimating counterfactual outcomes with spatial-temporal attributes using the Transformer, exhibiting stronger estimation ability. Under mild assumptions, the proposed estimator within this framework is consistent and asymptotically normal. To validate the effectiveness of our approach, we conduct simulation experiments and real data experiments. Simulation experiments show that our estimator has a stronger estimation capability than baseline methods. Real data experiments provide a valuable conclusion to the causal effect of conflicts on forest loss in Colombia. The source code is available at this URL.

Cite this Paper

BibTeX
@InProceedings{pmlr-v267-li25g, title = {Transformer-Based Spatial-Temporal Counterfactual Outcomes Estimation}, author = {Li, He and Chi, Haoang and Liu, Mingyu and Huang, Wanrong and Xu, Liyang and Yang, Wenjing}, booktitle = {Proceedings of the 42nd International Conference on Machine Learning}, pages = {34185--34208}, year = {2025}, editor = {Singh, Aarti and Fazel, Maryam and Hsu, Daniel and Lacoste-Julien, Simon and Berkenkamp, Felix and Maharaj, Tegan and Wagstaff, Kiri and Zhu, Jerry}, volume = {267}, series = {Proceedings of Machine Learning Research}, month = {13--19 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v267/main/assets/li25g/li25g.pdf}, url = {https://proceedings.mlr.press/v267/li25g.html}, abstract = {The real world naturally has dimensions of time and space. Therefore, estimating the counterfactual outcomes with spatial-temporal attributes is a crucial problem. However, previous methods are based on classical statistical models, which still have limitations in performance and generalization. This paper proposes a novel framework for estimating counterfactual outcomes with spatial-temporal attributes using the Transformer, exhibiting stronger estimation ability. Under mild assumptions, the proposed estimator within this framework is consistent and asymptotically normal. To validate the effectiveness of our approach, we conduct simulation experiments and real data experiments. Simulation experiments show that our estimator has a stronger estimation capability than baseline methods. Real data experiments provide a valuable conclusion to the causal effect of conflicts on forest loss in Colombia. The source code is available at this URL.} }
Endnote
%0 Conference Paper %T Transformer-Based Spatial-Temporal Counterfactual Outcomes Estimation %A He Li %A Haoang Chi %A Mingyu Liu %A Wanrong Huang %A Liyang Xu %A Wenjing Yang %B Proceedings of the 42nd International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2025 %E Aarti Singh %E Maryam Fazel %E Daniel Hsu %E Simon Lacoste-Julien %E Felix Berkenkamp %E Tegan Maharaj %E Kiri Wagstaff %E Jerry Zhu %F pmlr-v267-li25g %I PMLR %P 34185--34208 %U https://proceedings.mlr.press/v267/li25g.html %V 267 %X The real world naturally has dimensions of time and space. Therefore, estimating the counterfactual outcomes with spatial-temporal attributes is a crucial problem. However, previous methods are based on classical statistical models, which still have limitations in performance and generalization. This paper proposes a novel framework for estimating counterfactual outcomes with spatial-temporal attributes using the Transformer, exhibiting stronger estimation ability. Under mild assumptions, the proposed estimator within this framework is consistent and asymptotically normal. To validate the effectiveness of our approach, we conduct simulation experiments and real data experiments. Simulation experiments show that our estimator has a stronger estimation capability than baseline methods. Real data experiments provide a valuable conclusion to the causal effect of conflicts on forest loss in Colombia. The source code is available at this URL.
APA
Li, H., Chi, H., Liu, M., Huang, W., Xu, L. & Yang, W.. (2025). Transformer-Based Spatial-Temporal Counterfactual Outcomes Estimation. Proceedings of the 42nd International Conference on Machine Learning, in Proceedings of Machine Learning Research 267:34185-34208 Available from https://proceedings.mlr.press/v267/li25g.html.

Related Material