EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

AALF: Almost Always Linear Forecasting

Matthias Jakobs, Thomas Liebig

Machine Learning · 2026-03 · 115 (3)

DOI: 10.1007/s10994-026-07020-2 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Machine Learning, online erschienen am 06.03.2026. Verfasst von Matthias Jakobs und 1 weiteren Autor:innen. Die Arbeit wurde bisher 1-mal zitiert (Crossref) und nennt 38 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Technische Universität Dortmund.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Recent work for time-series forecasting increasingly leverages the high predictive power of Deep Learning models. With this increase in model complexity, however, comes a lack in understanding of the underlying model decision process, which is problematic for safety-critical application scenarios. At the same time, simple, interpretable forecasting methods such as ARIMA and ETS still perform very well, sometimes on-par with Deep Learning approaches. We argue that using interpretable forecasters leads to good predictions in most cases. However, the forecasting performance can be improved by selecting a Deep Learning method only for few, important predictions, increasing the overall interpretability of the forecasting process. In this context, we propose a novel online model selection framework which learns to identify these predictions. An extensive empirical study on various real-world datasets containing over 3500 individual time-series shows that our selection methodology performs comparable to state-of-the-art online model selection methods in most cases while being significantly more interpretable. We find that almost always choosing a simple autoregressive or exponential smoothing model for forecasting, results in competitive performance, suggesting that the need for opaque black-box models in time-series forecasting might be smaller than recent works would suggest.

Förderung

  • Technische Universität Dortmund

Zitieren

APA

Jakobs, M., & Liebig, T. (2026). AALF: Almost Always Linear Forecasting. Machine Learning, 115(3). https://doi.org/10.1007/s10994-026-07020-2

Vancouver

Jakobs M, Liebig T. AALF: Almost Always Linear Forecasting. Machine Learning. 2026;115(3). doi:10.1007/s10994-026-07020-2

BibTeX
@article{Jakobs2026,
  author = {Jakobs, Matthias and Liebig, Thomas},
  title = {AALF: Almost Always Linear Forecasting},
  journal = {Machine Learning},
  year = {2026},
  volume = {115},
  number = {3},
  doi = {10.1007/s10994-026-07020-2}
}
Kostenlos starten →