Online Change-Point Detection in Vector Autoregressive Models: A Score-Driven Bayesian Approach
Entropy · 07.11.2026 · 28 (11) · 1227
DOI: 10.3390/e28111227 ↗Kurz gefasst
Fachartikel in der Zeitschrift Entropy, online erschienen am 07.11.2026. Verfasst von Xueqing Zhang und 1 weiteren Autor:innen. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: National Natural Science Foundation of China, Natural Science Foundation of Qinghai Province.
Abstract (Originalsprache)
This paper proposes a score-driven Bayesian method to detect online mean change-points in a vector autoregressive model. The within-state process follows a multivariate autoregressive structure with time-varying autoregressive coefficient matrices and a time-varying conditional covariance matrix. The time-varying parameters follow a generalized autoregressive score updating mechanism. Under Gaussian assumptions, we derive a closed-form expression for the underlying predictive model and recursive update equations for the posterior hyperparameters. A two-layer inference architecture integrates particle filtering with sequential maximum likelihood estimation. Four order-identification schemes are proposed. Monte Carlo experiments show that the method outperforms component-wise detection and fixed-parameter models in predictive accuracy and detection coverage. Empirical analyses of U.S. macroeconomic indicators and global stock indices validate the practicality of the proposed method.
Förderung
- National Natural Science Foundation of China
- Natural Science Foundation of Qinghai Province
Zitieren
APA
Zhang, X., & Chen, Z. (2026). Online Change-Point Detection in Vector Autoregressive Models: A Score-Driven Bayesian Approach. Entropy, 28(11), 1227. https://doi.org/10.3390/e28111227
Vancouver
Zhang X, Chen Z. Online Change-Point Detection in Vector Autoregressive Models: A Score-Driven Bayesian Approach. Entropy. 2026;28(11):1227. doi:10.3390/e28111227
BibTeX
@article{Zhang2026,
author = {Zhang, Xueqing and Chen, Zhanshou},
title = {Online Change-Point Detection in Vector Autoregressive Models: A Score-Driven Bayesian Approach},
journal = {Entropy},
year = {2026},
volume = {28},
number = {11},
pages = {1227},
doi = {10.3390/e28111227}
}
EUVIMED