Hierarchies Everywhere -- Managing & Measuring Uncertainty in Hierarchical Time Series

TL;DR

This paper introduces a Bayesian hierarchical approach to improve the accuracy and interpretability of large, connected time series forecasts by exploiting their structure and reducing dimensionality.

Contribution

It presents a novel Bayesian framework that explicitly models hierarchical connectedness, enhances forecast accuracy, and allows assessment of value added at each level.

Findings

Improved forecast accuracy through hierarchical Bayesian modeling

Ability to assess forecast value added at each hierarchy level

Probabilistic forecasts derived from the approach

Abstract

We examine the problem of making reconciled forecasts of large collections of related time series through a behavioural/Bayesian lens. Our approach explicitly acknowledges and exploits the 'connectedness' of the series in terms of time-series characteristics and forecast accuracy as well as hierarchical structure. By making maximal use of the available information, and by significantly reducing the dimensionality of the hierarchical forecasting problem, we show how to improve the accuracy of the reconciled forecasts. In contrast to existing approaches, our structure allows the analysis and assessment of the forecast value added at each hierarchical level. Our reconciled forecasts are inherently probabilistic, whether probabilistic base forecasts are used or not.