Coherent forecast combination for linearly constrained multiple time series

TL;DR

This paper introduces an optimization-based method for combining multiple unbiased forecasts of linearly constrained multivariate time series, ensuring coherence and improving accuracy in practical hierarchical forecasting contexts.

Contribution

It develops a closed-form solution for coherent forecast combination that respects linear constraints and addresses covariance matrix issues, advancing hierarchical forecast reconciliation methods.

Findings

Significant forecast accuracy improvements demonstrated in simulations.

Method ensures coherence while combining multiple forecasts.

Outperforms single-task and single-expert approaches in experiments.

Abstract

Linearly constrained multiple time series may be encountered in many practical contexts, such as the National Accounts (e.g., GDP disaggregated by Income, Expenditure and Output), and multilevel frameworks where the variables are organized according to hierarchies or groupings, like the total energy consumption of a country disaggregated by region and energy sources. In these cases, when multiple incoherent base forecasts for each individual variable are available, a forecast combination-and-reconciliation approach, that we call coherent forecast combination, may be used to improve the accuracy of the base forecasts and achieve coherence in the final result. In this paper, we develop an optimization-based technique that combines multiple unbiased base forecasts while assuring the constraints valid for the series. We present closed form expressions for the coherent combined forecast vector and its error covariance matrix in the general case where a different number of forecasts is available for each variable. We also discuss practical issues related to the covariance matrix that is part of the optimal solution. Through simulations and a forecasting experiment on the daily Australian electricity generation hierarchical time series, we show that the proposed methodology, in addition to adhering to sound statistical principles, may yield in significant improvement on base forecasts, single-task combination and single-expert reconciliation approaches as well.