Reconciling Hierarchical Forecasts via Bayes' Rule
Applications
2021-06-07 v5
Abstract
We present a novel approach for reconciling hierarchical forecasts, based on Bayes rule. We define a prior distribution for the bottom time series of the hierarchy, based on the bottom base forecasts. Then we update their distribution via Bayes rule, based on the base forecasts for the upper time series. Under the Gaussian assumption, we derive the updating in closed-form. We derive two algorithms, which differ as for the assumed independencies. We discuss their relation with the MinT reconciliation algorithm and with the Kalman filter, and we compare them experimentally.
Cite
@article{arxiv.1906.03105,
title = {Reconciling Hierarchical Forecasts via Bayes' Rule},
author = {Giorgio Corani and Dario Azzimonti and João P. S. C. Augusto and Marco Zaffalon},
journal= {arXiv preprint arXiv:1906.03105},
year = {2021}
}