Causal Search in Structural Vector Autoregressive Models

Alessio Moneta, Nadine Chlass, Doris Entner, Patrik Hoyer
Proceedings of the Neural Information Processing Systems Mini-Symposium on Causality in Time Series, PMLR 12:95-114, 2011.

Abstract

This paper reviews a class of methods to perform causal inference in the framework of a structural vector autoregressive model. We consider three different settings. In the first setting the underlying system is linear with normal disturbances and the structural model is identified by exploiting the information incorporated in the partial correlations of the estimated residuals. Zero partial correlations are used as input of a search algorithm formalized via graphical causal models. In the second, semi-parametric, setting the underlying system is linear with non-Gaussian disturbances. In this case the structural vector autoregressive model is identified through a search procedure based on independent component analysis. Finally, we explore the possibility of causal search in a nonparametric setting by studying the performance of conditional independence tests based on kernel density estimations.

Cite this Paper

BibTeX
@InProceedings{pmlr-v12-moneta11, title = {Causal Search in Structural Vector Autoregressive Models}, author = {Moneta, Alessio and Chlass, Nadine and Entner, Doris and Hoyer, Patrik}, booktitle = {Proceedings of the Neural Information Processing Systems Mini-Symposium on Causality in Time Series}, pages = {95--114}, year = {2011}, editor = {Popescu, Florin and Guyon, Isabelle}, volume = {12}, series = {Proceedings of Machine Learning Research}, address = {Vancouver, Canada}, month = {10 Dec}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v12/moneta11/moneta11.pdf}, url = {https://proceedings.mlr.press/v12/moneta11.html}, abstract = {This paper reviews a class of methods to perform causal inference in the framework of a structural vector autoregressive model. We consider three different settings. In the first setting the underlying system is linear with normal disturbances and the structural model is identified by exploiting the information incorporated in the partial correlations of the estimated residuals. Zero partial correlations are used as input of a search algorithm formalized via graphical causal models. In the second, semi-parametric, setting the underlying system is linear with non-Gaussian disturbances. In this case the structural vector autoregressive model is identified through a search procedure based on independent component analysis. Finally, we explore the possibility of causal search in a nonparametric setting by studying the performance of conditional independence tests based on kernel density estimations.} }
Endnote
%0 Conference Paper %T Causal Search in Structural Vector Autoregressive Models %A Alessio Moneta %A Nadine Chlass %A Doris Entner %A Patrik Hoyer %B Proceedings of the Neural Information Processing Systems Mini-Symposium on Causality in Time Series %C Proceedings of Machine Learning Research %D 2011 %E Florin Popescu %E Isabelle Guyon %F pmlr-v12-moneta11 %I PMLR %P 95--114 %U https://proceedings.mlr.press/v12/moneta11.html %V 12 %X This paper reviews a class of methods to perform causal inference in the framework of a structural vector autoregressive model. We consider three different settings. In the first setting the underlying system is linear with normal disturbances and the structural model is identified by exploiting the information incorporated in the partial correlations of the estimated residuals. Zero partial correlations are used as input of a search algorithm formalized via graphical causal models. In the second, semi-parametric, setting the underlying system is linear with non-Gaussian disturbances. In this case the structural vector autoregressive model is identified through a search procedure based on independent component analysis. Finally, we explore the possibility of causal search in a nonparametric setting by studying the performance of conditional independence tests based on kernel density estimations.
RIS
TY - CPAPER TI - Causal Search in Structural Vector Autoregressive Models AU - Alessio Moneta AU - Nadine Chlass AU - Doris Entner AU - Patrik Hoyer BT - Proceedings of the Neural Information Processing Systems Mini-Symposium on Causality in Time Series DA - 2011/03/03 ED - Florin Popescu ED - Isabelle Guyon ID - pmlr-v12-moneta11 PB - PMLR DP - Proceedings of Machine Learning Research VL - 12 SP - 95 EP - 114 L1 - http://proceedings.mlr.press/v12/moneta11/moneta11.pdf UR - https://proceedings.mlr.press/v12/moneta11.html AB - This paper reviews a class of methods to perform causal inference in the framework of a structural vector autoregressive model. We consider three different settings. In the first setting the underlying system is linear with normal disturbances and the structural model is identified by exploiting the information incorporated in the partial correlations of the estimated residuals. Zero partial correlations are used as input of a search algorithm formalized via graphical causal models. In the second, semi-parametric, setting the underlying system is linear with non-Gaussian disturbances. In this case the structural vector autoregressive model is identified through a search procedure based on independent component analysis. Finally, we explore the possibility of causal search in a nonparametric setting by studying the performance of conditional independence tests based on kernel density estimations. ER -
APA
Moneta, A., Chlass, N., Entner, D. & Hoyer, P.. (2011). Causal Search in Structural Vector Autoregressive Models. Proceedings of the Neural Information Processing Systems Mini-Symposium on Causality in Time Series, in Proceedings of Machine Learning Research 12:95-114 Available from https://proceedings.mlr.press/v12/moneta11.html.

Related Material