Online aggregation of conformal predictive systems

Vladimir G. Trunov, Vladimir V. V’yugin
Proceedings of the Twelfth Symposium on Conformal and Probabilistic Prediction with Applications, PMLR 204:430-449, 2023.

Abstract

The problem of online probabilistic forecasting is considered. Probabilistic forecasts are obtained as a result of the application of conformal predictive systems. The conformal predictive system is a novel method for obtaining reliable predictions which are based on point forecasts of the regression algorithm. The paper considers the case when at each moment of time several competing conformal predictive systems (experts) give their predictions in the form of probability distribution functions. Probabilistic forecasts of the experts are combined by an aggregation algorithm into one probabilistic forecast at each step of the forecasting process, while expert forecasts can be used partially. The developed methods are used to solve the well-known problem of predicting the load of an electrical network online. Numerical experiments have shown the agreement of predictions with real data.

Cite this Paper

BibTeX
@InProceedings{pmlr-v204-trunov23a, title = {Online aggregation of conformal predictive systems}, author = {Trunov, Vladimir G. and {V'yugin}, Vladimir V.}, booktitle = {Proceedings of the Twelfth Symposium on Conformal and Probabilistic Prediction with Applications}, pages = {430--449}, year = {2023}, editor = {Papadopoulos, Harris and Nguyen, Khuong An and Boström, Henrik and Carlsson, Lars}, volume = {204}, series = {Proceedings of Machine Learning Research}, month = {13--15 Sep}, publisher = {PMLR}, pdf = {https://proceedings.mlr.press/v204/trunov23a/trunov23a.pdf}, url = {https://proceedings.mlr.press/v204/trunov23a.html}, abstract = {The problem of online probabilistic forecasting is considered. Probabilistic forecasts are obtained as a result of the application of conformal predictive systems. The conformal predictive system is a novel method for obtaining reliable predictions which are based on point forecasts of the regression algorithm. The paper considers the case when at each moment of time several competing conformal predictive systems (experts) give their predictions in the form of probability distribution functions. Probabilistic forecasts of the experts are combined by an aggregation algorithm into one probabilistic forecast at each step of the forecasting process, while expert forecasts can be used partially. The developed methods are used to solve the well-known problem of predicting the load of an electrical network online. Numerical experiments have shown the agreement of predictions with real data.} }
Endnote
%0 Conference Paper %T Online aggregation of conformal predictive systems %A Vladimir G. Trunov %A Vladimir V. V’yugin %B Proceedings of the Twelfth Symposium on Conformal and Probabilistic Prediction with Applications %C Proceedings of Machine Learning Research %D 2023 %E Harris Papadopoulos %E Khuong An Nguyen %E Henrik Boström %E Lars Carlsson %F pmlr-v204-trunov23a %I PMLR %P 430--449 %U https://proceedings.mlr.press/v204/trunov23a.html %V 204 %X The problem of online probabilistic forecasting is considered. Probabilistic forecasts are obtained as a result of the application of conformal predictive systems. The conformal predictive system is a novel method for obtaining reliable predictions which are based on point forecasts of the regression algorithm. The paper considers the case when at each moment of time several competing conformal predictive systems (experts) give their predictions in the form of probability distribution functions. Probabilistic forecasts of the experts are combined by an aggregation algorithm into one probabilistic forecast at each step of the forecasting process, while expert forecasts can be used partially. The developed methods are used to solve the well-known problem of predicting the load of an electrical network online. Numerical experiments have shown the agreement of predictions with real data.
APA
Trunov, V.G. & V’yugin, V.V.. (2023). Online aggregation of conformal predictive systems. Proceedings of the Twelfth Symposium on Conformal and Probabilistic Prediction with Applications, in Proceedings of Machine Learning Research 204:430-449 Available from https://proceedings.mlr.press/v204/trunov23a.html.

Related Material