Interpreting Primary Energy Consumption in Europe and Türkiye Using Explainable Artificial Intelligence

Abstract

This study aims to analyze the key determinants of primary energy consumption in Europe and Türkiye by interpreting the results obtained through machine learning (ML) methods using explainable artificial intelligence (XAI) techniques. The analysis was conducted separately for European countries and Türkiye, using annual energy, economic, and demographic data covering the period from 1965 to 2022. The model was constructed using a regression-based ML framework, and SHapley Additive exPlanations (SHAP) and local interpretable model– agnostic explanation (LIME) methods were used to ensure the interpretability of model outputs both globally and locally. The model achieved high predictive performance with an R2 of 0.988, a root mean squared error of 97.818, a mean absolute error of 69.55, a normalized root mean squared error of 0.022, and a normalized mean absolute error of 0.016, normalized using the min–max method. The results show that oil consumption and gross domestic product (GDP) continue to be the dominant drivers of primary energy demand across Europe, while the share of renewable energy exhibits a suppressive effect. Comparative SHAP and LIME analyses for Türkiye reveal that, while the share of renewable energy is beginning to negatively impact energy consumption, fossil fuel dependence, and GDP-driven demand remain dominant. The findings provide directly interpretable, policy-relevant insights for sustainable energy planning and demonstrate the added value of XAI-based approaches to energy consumption analysis.

Cite this article as: A. N. Adıgüzel Tüylü, “Interpreting primary energy consumption in Europe and Türkiye using explainable artificial intelligence,” Electrica, 2026, 26, 0405, doi: 10.5152/electr.2026.25405.