Forecasting Multivariate Time Series under Predictive Heterogeneity: A Validation-Driven Clustering Framework

Ziling Ma 1 , Ángel López-Oriona 1 , Hernando Ombao 1 , Ying Sun 1 · 2026

Read Forecasting Multivariate Time Series under Predictive Heterogeneity: A Validation-Driven Clustering Framework by Ziling Ma 1 , Ángel López-Oriona 1 , Hernando Ombao 1 , Ying Sun 1 (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2604-13748.txt.

Description

Ma and three colleagues forecast many series over time, where each series behaves differently. One model for all can miss the differences. A model per series can go wrong. Their framework groups series by out-of-sample predictive performance, not by similar representations. Assignments update from validation losses for point and probabilistic forecasts. A fallback returns to the global model when specialization fails to help. Traffic tests show gains without degradation when differences are weak.

Catalog record

CREATORZiling Ma 1 , Ángel López-Oriona 1 , Hernando Ombao 1 , Ying Sun 1
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE68894
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2604.13748

Browse the full library · About the Tower of Babel