How to use forecast reconciliation: Cross-temporal probabilistic forecast reconciliation
Abstract
Forecast reconciliation is a post-forecasting process intended to improve the quality of forecasts for a system of linearly constrained multiple time series. It is valuable in various fields like GDP components, electricity demand, supply chain demand with product categories, and tourist arrivals. Moreover, effective decision-making depends on the support of accurate and coherent forecasts.
In this talk, we will explore point and probabilistic forecast reconciliation within the broader cross-temporal framework. We’ll delve into how these methods can enhance forecast accuracy and support better decision-making. Additionally, we’ll demonstrate how to implement and apply these techniques using the R package FoReco.