Forecasting electricity prices with machine learning: predictor sensitivity

Abstract

Purpose

Trading on electricity markets occurs such that the price settlement takes place before delivery, often day-ahead. In practice, these prices are highly volatile as they largely depend upon a range of variables such as electricity demand and the feed-in from renewable energy sources. Hence, the purpose of this paper is to provide accurate forecasts..

Design/methodology/approach

This paper aims at comparing different predictors stemming from supply-side (solar and wind power generation), demand-side, fuel-related and economic influences. For this reason, this paper implements a broad range of non-linear models from machine learning and draw upon the information-fusion-based sensitivity analysis.

Findings

This study disentangles the respective relevance of each predictor. This study shows that external predictors altogether decrease root mean squared errors by up to 21.96%. A Diebold-Mariano test statistically proves that the forecasting accuracy of the proposed machine learning models is superior.

Research limitations/implications

The performance gain from including more predictors might be larger than from a better model. Future research should place attention on expanding the data basis in electricity price forecasting.

Practical implications

When developing pricing models, practitioners can achieve reasonable performance with a simple model (e.g. seasonal-autoregressive moving-average) that is built upon a wide range of predictors.

Originality/value

The benefit of adding further predictors has only recently received traction; however, little is known about how the individual variables contribute to improving forecasts in machine learning.

Citation

Naumzik, C. and Feuerriegel, S. (2021), "Forecasting electricity prices with machine learning: predictor sensitivity", International Journal of Energy Sector Management, Vol. 15 No. 1, pp. 157-172. https://doi.org/10.1108/IJESM-01-2020-0001