An explainable artificial intelligence feature selection framework for transparent, trustworthy, and cost-efficient energy forecasting

Research output: Contribution to journalArticleResearchpeer review

Authors

View graph of relations

Details

Original languageEnglish
Article number100648
JournalEnergy and AI
Volume22
Early online date25 Nov 2025
Publication statusPublished - Dec 2025

Abstract

Accurate forecasting of renewable power generation is crucial for grid stability and cost efficiency. Feature selection in AI-based forecasting remains challenging due to high data acquisition cost, lack of transparency, and limited user control. We introduce a transparent and cost-sensitive feature selection framework for renewable power forecasting that leverages Explainable Artificial Intelligence (XAI). We integrate SHapley Additive exPlanations (SHAP) and Explain Like I'm 5 (ELI5) to identify dominant and redundant features. This approach enables systematic dataset reduction without compromising model performance. Our case study, based on Photovoltaic (PV) generation data, evaluates the approach across four experimental setups. Experimental results indicate that our XAI-based feature selection reduces the dominance index from 0.37 to 0.17, maintains high predictive accuracy (R2 = 0.94, drop < 0.04), and lowers data acquisition costs. Furthermore, eliminating dominant features improves robustness to noise and reduces performance variance by a factor of three compared to the baseline scenario. The developed framework enhances interpretability, supports human-in-the-loop decision-making, and introduces a cost-sensitive objective function for feature selection. By combining transparency, robustness, and efficiency, we contribute to the development and implementation of Trustworthy AI (TAI) applications in energy forecasting, providing a scalable solution for industrial deployment.

Keywords

    Cost efficiency, Energy sector, Explainable artificial intelligence, Feature reduction, Robustness, XAI-feature selection Framework

ASJC Scopus subject areas

Sustainable Development Goals

Cite this

An explainable artificial intelligence feature selection framework for transparent, trustworthy, and cost-efficient energy forecasting. / Kost, Leonard; Lier, Sarah K.; Breitner, Michael H.
In: Energy and AI, Vol. 22, 100648, 12.2025.

Research output: Contribution to journalArticleResearchpeer review

Kost L, Lier SK, Breitner MH. An explainable artificial intelligence feature selection framework for transparent, trustworthy, and cost-efficient energy forecasting. Energy and AI. 2025 Dec;22:100648. Epub 2025 Nov 25. doi: 10.1016/j.egyai.2025.100648
Download
@article{4a01d4ad7c044ef58de4380d62d98f99,
title = "An explainable artificial intelligence feature selection framework for transparent, trustworthy, and cost-efficient energy forecasting",
abstract = "Accurate forecasting of renewable power generation is crucial for grid stability and cost efficiency. Feature selection in AI-based forecasting remains challenging due to high data acquisition cost, lack of transparency, and limited user control. We introduce a transparent and cost-sensitive feature selection framework for renewable power forecasting that leverages Explainable Artificial Intelligence (XAI). We integrate SHapley Additive exPlanations (SHAP) and Explain Like I'm 5 (ELI5) to identify dominant and redundant features. This approach enables systematic dataset reduction without compromising model performance. Our case study, based on Photovoltaic (PV) generation data, evaluates the approach across four experimental setups. Experimental results indicate that our XAI-based feature selection reduces the dominance index from 0.37 to 0.17, maintains high predictive accuracy (R2 = 0.94, drop < 0.04), and lowers data acquisition costs. Furthermore, eliminating dominant features improves robustness to noise and reduces performance variance by a factor of three compared to the baseline scenario. The developed framework enhances interpretability, supports human-in-the-loop decision-making, and introduces a cost-sensitive objective function for feature selection. By combining transparency, robustness, and efficiency, we contribute to the development and implementation of Trustworthy AI (TAI) applications in energy forecasting, providing a scalable solution for industrial deployment.",
keywords = "Cost efficiency, Energy sector, Explainable artificial intelligence, Feature reduction, Robustness, XAI-feature selection Framework",
author = "Leonard Kost and Lier, {Sarah K.} and Breitner, {Michael H.}",
note = "Publisher Copyright: {\textcopyright} 2025 The Authors",
year = "2025",
month = dec,
doi = "10.1016/j.egyai.2025.100648",
language = "English",
volume = "22",

}

Download

TY - JOUR

T1 - An explainable artificial intelligence feature selection framework for transparent, trustworthy, and cost-efficient energy forecasting

AU - Kost, Leonard

AU - Lier, Sarah K.

AU - Breitner, Michael H.

N1 - Publisher Copyright: © 2025 The Authors

PY - 2025/12

Y1 - 2025/12

N2 - Accurate forecasting of renewable power generation is crucial for grid stability and cost efficiency. Feature selection in AI-based forecasting remains challenging due to high data acquisition cost, lack of transparency, and limited user control. We introduce a transparent and cost-sensitive feature selection framework for renewable power forecasting that leverages Explainable Artificial Intelligence (XAI). We integrate SHapley Additive exPlanations (SHAP) and Explain Like I'm 5 (ELI5) to identify dominant and redundant features. This approach enables systematic dataset reduction without compromising model performance. Our case study, based on Photovoltaic (PV) generation data, evaluates the approach across four experimental setups. Experimental results indicate that our XAI-based feature selection reduces the dominance index from 0.37 to 0.17, maintains high predictive accuracy (R2 = 0.94, drop < 0.04), and lowers data acquisition costs. Furthermore, eliminating dominant features improves robustness to noise and reduces performance variance by a factor of three compared to the baseline scenario. The developed framework enhances interpretability, supports human-in-the-loop decision-making, and introduces a cost-sensitive objective function for feature selection. By combining transparency, robustness, and efficiency, we contribute to the development and implementation of Trustworthy AI (TAI) applications in energy forecasting, providing a scalable solution for industrial deployment.

AB - Accurate forecasting of renewable power generation is crucial for grid stability and cost efficiency. Feature selection in AI-based forecasting remains challenging due to high data acquisition cost, lack of transparency, and limited user control. We introduce a transparent and cost-sensitive feature selection framework for renewable power forecasting that leverages Explainable Artificial Intelligence (XAI). We integrate SHapley Additive exPlanations (SHAP) and Explain Like I'm 5 (ELI5) to identify dominant and redundant features. This approach enables systematic dataset reduction without compromising model performance. Our case study, based on Photovoltaic (PV) generation data, evaluates the approach across four experimental setups. Experimental results indicate that our XAI-based feature selection reduces the dominance index from 0.37 to 0.17, maintains high predictive accuracy (R2 = 0.94, drop < 0.04), and lowers data acquisition costs. Furthermore, eliminating dominant features improves robustness to noise and reduces performance variance by a factor of three compared to the baseline scenario. The developed framework enhances interpretability, supports human-in-the-loop decision-making, and introduces a cost-sensitive objective function for feature selection. By combining transparency, robustness, and efficiency, we contribute to the development and implementation of Trustworthy AI (TAI) applications in energy forecasting, providing a scalable solution for industrial deployment.

KW - Cost efficiency

KW - Energy sector

KW - Explainable artificial intelligence

KW - Feature reduction

KW - Robustness

KW - XAI-feature selection Framework

UR - http://www.scopus.com/inward/record.url?scp=105023096365&partnerID=8YFLogxK

U2 - 10.1016/j.egyai.2025.100648

DO - 10.1016/j.egyai.2025.100648

M3 - Article

AN - SCOPUS:105023096365

VL - 22

JO - Energy and AI

JF - Energy and AI

M1 - 100648

ER -

By the same author(s)