Conformal Multistep-Ahead Multivariate Time-Series Forecasting

Filip Schlembach, Evgueni Smirnov, Irena Koprinska
Proceedings of the Eleventh Symposium on Conformal and Probabilistic Prediction with Applications, PMLR 179:316-318, 2022.

Abstract

This paper proposes a method for conformal multistep-ahead multivariate time-series forecasting. The method minimizes the coverage loss when the data exchangeability assumption does not properly hold. This is done by weighting residual quantiles while computing prediction intervals. Preliminary experiments on real data demonstrate the method’s utility.

Cite this Paper


BibTeX
@InProceedings{pmlr-v179-schlembach22a, title = {Conformal Multistep-Ahead Multivariate Time-Series Forecasting}, author = {Schlembach, Filip and Smirnov, Evgueni and Koprinska, Irena}, booktitle = {Proceedings of the Eleventh Symposium on Conformal and Probabilistic Prediction with Applications}, pages = {316--318}, year = {2022}, editor = {Johansson, Ulf and Boström, Henrik and An Nguyen, Khuong and Luo, Zhiyuan and Carlsson, Lars}, volume = {179}, series = {Proceedings of Machine Learning Research}, month = {24--26 Aug}, publisher = {PMLR}, pdf = {https://proceedings.mlr.press/v179/schlembach22a/schlembach22a.pdf}, url = {https://proceedings.mlr.press/v179/schlembach22a.html}, abstract = {This paper proposes a method for conformal multistep-ahead multivariate time-series forecasting. The method minimizes the coverage loss when the data exchangeability assumption does not properly hold. This is done by weighting residual quantiles while computing prediction intervals. Preliminary experiments on real data demonstrate the method’s utility.} }
Endnote
%0 Conference Paper %T Conformal Multistep-Ahead Multivariate Time-Series Forecasting %A Filip Schlembach %A Evgueni Smirnov %A Irena Koprinska %B Proceedings of the Eleventh Symposium on Conformal and Probabilistic Prediction with Applications %C Proceedings of Machine Learning Research %D 2022 %E Ulf Johansson %E Henrik Boström %E Khuong An Nguyen %E Zhiyuan Luo %E Lars Carlsson %F pmlr-v179-schlembach22a %I PMLR %P 316--318 %U https://proceedings.mlr.press/v179/schlembach22a.html %V 179 %X This paper proposes a method for conformal multistep-ahead multivariate time-series forecasting. The method minimizes the coverage loss when the data exchangeability assumption does not properly hold. This is done by weighting residual quantiles while computing prediction intervals. Preliminary experiments on real data demonstrate the method’s utility.
APA
Schlembach, F., Smirnov, E. & Koprinska, I.. (2022). Conformal Multistep-Ahead Multivariate Time-Series Forecasting. Proceedings of the Eleventh Symposium on Conformal and Probabilistic Prediction with Applications, in Proceedings of Machine Learning Research 179:316-318 Available from https://proceedings.mlr.press/v179/schlembach22a.html.

Related Material