A Context-Dependent Method for Adaptive Tuning of Parameters of Autoregressive Models for Non-Stationary Time Series
Informatics and Automation · 06.04.2026 · 25 (2) · 352-377
DOI: 10.15622/ia.25.2.4 ↗Kurz gefasst
Fachartikel in der Zeitschrift Informatics and Automation, online erschienen am 06.04.2026. Verfasst von Andrei Vorobev und 1 weiteren Autor:innen. Die Arbeit nennt 21 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).
Abstract (Originalsprache)
A method for context-sensitive tuning of autoregressive model parameters for gap reconstruction in nonstationary time series is proposed. A key feature of the method is the adaptive selection of ARIMA (p, d, q) model parameters based on two context factors: the gap duration and the level of external disturbances during the corresponding period. Unlike standard automatic model selection approaches focused on global optimization for forecasting, the developed algorithm narrows the model search space and selects the optimal configuration using local cross-validation, allowing for consideration of specific conditions in the gap region. The method is implemented as a Python software module with a modular architecture that ensures computational efficiency through caching and parallel computing. The effectiveness of the method was tested experimentally on real geomagnetic data (the DBE_NEZ component of the Lovozero Observatory). The results demonstrate that under calm and weakly disturbed geomagnetic conditions (SME index = 50–200 nT), the method provides high reconstruction accuracy (R² = 0.71–0.85) for gaps ranging from 5 to 120 minutes in length. However, accuracy is shown to decrease consistently with increasing disturbance level, reflecting a fundamental limitation associated with the increasing stochasticity of the original signal. The proposed approach ensures interpretability and adaptability, opening up prospects for the development of data reconstruction tools in various application areas.
Zitieren
APA
Vorobev, A., & Vorobeva, G. (2026). A Context-Dependent Method for Adaptive Tuning of Parameters of Autoregressive Models for Non-Stationary Time Series. Informatics and Automation, 25(2), 352-377. https://doi.org/10.15622/ia.25.2.4
Vancouver
Vorobev A, Vorobeva G. A Context-Dependent Method for Adaptive Tuning of Parameters of Autoregressive Models for Non-Stationary Time Series. Informatics and Automation. 2026;25(2):352-377. doi:10.15622/ia.25.2.4
BibTeX
@article{Vorobev2026,
author = {Vorobev, Andrei and Vorobeva, Gulnara},
title = {A Context-Dependent Method for Adaptive Tuning of Parameters of Autoregressive Models for Non-Stationary Time Series},
journal = {Informatics and Automation},
year = {2026},
volume = {25},
number = {2},
pages = {352--377},
doi = {10.15622/ia.25.2.4}
}
EUVIMED