Electricity Price Forecasting: Bridging Linear Models, Neural Networks and Online Learning

View PDF HTML (experimental)

Abstract:Precise day-ahead forecasts for electricity prices are crucial to ensure efficient portfolio management, support strategic decision-making for power plant operations, enable efficient battery storage optimization, and facilitate demand response planning. However, developing an accurate prediction model is highly challenging in an uncertain and volatile market environment. For instance, although linear models generally exhibit competitive performance in predicting electricity prices with minimal computational requirements, they fail to capture relevant nonlinear relationships. Nonlinear models, on the other hand, can improve forecasting accuracy with a surge in computational costs. We introduce a novel partial online learning approach, the key contribution of this work, which substantially reduces computational time. In addition, we propose a multivariate hybrid neural architecture that combines linear and nonlinear feed-forward neural structures. Unlike previous hybrid models, our approach integrates forecast combination using Bernstein Online Aggregation (BOA) to further improve forecasting accuracy. Compared to the current state-of-the-art benchmark models, the proposed forecasting method significantly reduces computational cost while delivering superior forecasting accuracy (11-12% RMSE and 14-17% MAE reductions). Our results are derived from a six-year forecasting study conducted on major European electricity markets.

Subjects:

Machine Learning (cs.LG)

Cite as: arXiv:2601.02856 [cs.LG]

(or arXiv:2601.02856v3 [cs.LG] for this version)

https://doi.org/10.48550/arXiv.2601.02856

arXiv-issued DOI via DataCite

Submission history

From: Btissame El Mahtout [view email] [v1] Tue, 6 Jan 2026 09:35:02 UTC (910 KB) [v2] Wed, 25 Mar 2026 18:32:08 UTC (696 KB) [v3] Fri, 27 Mar 2026 21:14:42 UTC (696 KB)