Estimating Identifiable Causal Effects on Markov Equivalence Class through Double Machine Learning

Yonghan Jung, Jin Tian, Elias Bareinboim
Proceedings of the 38th International Conference on Machine Learning, PMLR 139:5168-5179, 2021.

Abstract

General methods have been developed for estimating causal effects from observational data under causal assumptions encoded in the form of a causal graph. Most of this literature assumes that the underlying causal graph is completely specified. However, only observational data is available in most practical settings, which means that one can learn at most a Markov equivalence class (MEC) of the underlying causal graph. In this paper, we study the problem of causal estimation from a MEC represented by a partial ancestral graph (PAG), which is learnable from observational data. We develop a general estimator for any identifiable causal effects in a PAG. The result fills a gap for an end-to-end solution to causal inference from observational data to effects estimation. Specifically, we develop a complete identification algorithm that derives an influence function for any identifiable causal effects from PAGs. We then construct a double/debiased machine learning (DML) estimator that is robust to model misspecification and biases in nuisance function estimation, permitting the use of modern machine learning techniques. Simulation results corroborate with the theory.

Cite this Paper

BibTeX
@InProceedings{pmlr-v139-jung21b, title = {Estimating Identifiable Causal Effects on Markov Equivalence Class through Double Machine Learning}, author = {Jung, Yonghan and Tian, Jin and Bareinboim, Elias}, booktitle = {Proceedings of the 38th International Conference on Machine Learning}, pages = {5168--5179}, year = {2021}, editor = {Meila, Marina and Zhang, Tong}, volume = {139}, series = {Proceedings of Machine Learning Research}, month = {18--24 Jul}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v139/jung21b/jung21b.pdf}, url = {https://proceedings.mlr.press/v139/jung21b.html}, abstract = {General methods have been developed for estimating causal effects from observational data under causal assumptions encoded in the form of a causal graph. Most of this literature assumes that the underlying causal graph is completely specified. However, only observational data is available in most practical settings, which means that one can learn at most a Markov equivalence class (MEC) of the underlying causal graph. In this paper, we study the problem of causal estimation from a MEC represented by a partial ancestral graph (PAG), which is learnable from observational data. We develop a general estimator for any identifiable causal effects in a PAG. The result fills a gap for an end-to-end solution to causal inference from observational data to effects estimation. Specifically, we develop a complete identification algorithm that derives an influence function for any identifiable causal effects from PAGs. We then construct a double/debiased machine learning (DML) estimator that is robust to model misspecification and biases in nuisance function estimation, permitting the use of modern machine learning techniques. Simulation results corroborate with the theory.} }
Endnote
%0 Conference Paper %T Estimating Identifiable Causal Effects on Markov Equivalence Class through Double Machine Learning %A Yonghan Jung %A Jin Tian %A Elias Bareinboim %B Proceedings of the 38th International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2021 %E Marina Meila %E Tong Zhang %F pmlr-v139-jung21b %I PMLR %P 5168--5179 %U https://proceedings.mlr.press/v139/jung21b.html %V 139 %X General methods have been developed for estimating causal effects from observational data under causal assumptions encoded in the form of a causal graph. Most of this literature assumes that the underlying causal graph is completely specified. However, only observational data is available in most practical settings, which means that one can learn at most a Markov equivalence class (MEC) of the underlying causal graph. In this paper, we study the problem of causal estimation from a MEC represented by a partial ancestral graph (PAG), which is learnable from observational data. We develop a general estimator for any identifiable causal effects in a PAG. The result fills a gap for an end-to-end solution to causal inference from observational data to effects estimation. Specifically, we develop a complete identification algorithm that derives an influence function for any identifiable causal effects from PAGs. We then construct a double/debiased machine learning (DML) estimator that is robust to model misspecification and biases in nuisance function estimation, permitting the use of modern machine learning techniques. Simulation results corroborate with the theory.
APA
Jung, Y., Tian, J. & Bareinboim, E.. (2021). Estimating Identifiable Causal Effects on Markov Equivalence Class through Double Machine Learning. Proceedings of the 38th International Conference on Machine Learning, in Proceedings of Machine Learning Research 139:5168-5179 Available from https://proceedings.mlr.press/v139/jung21b.html.

Related Material