Integrating machine learning and econometric models to uncover macroeconomic determinants of renewable energy production in the selected European countries

The transition to renewable energy is a strategic priority across Europe, yet limited attention has been paid to the macroeconomic and institutional determinants of renewable energy production (REP). Understanding these factors is essential for crafting effective and resilient energy policies, particularly in a region characterized by diverse economic and governance contexts. This study investigates the drivers of REP in 26 European countries from 1995 to 2022, integrating panel econometric analysis with machine learning forecasting. Driscoll-Kraay Standard Errors (DKSE) and Fixed Effects (FE) models are compared, with DKSE preferred due to its ability to address heteroskedasticity, autocorrelation, and cross-sectional dependence. In addition, five machine learning models—including Random Forest, Support Vector Machine, CNN-BiLSTM-AR, LSTM, and ARIMA—are used to evaluate forecasting accuracy. The results identify research and development (R&D) expenditure as a dominant positive driver of REP, while political instability and weak rule of law significantly hinder progress. Macroeconomic variables such as GDP, inflation, population, and financial development also influence REP to varying degrees. Among forecasting models, Random Forest achieves the highest predictive accuracy across most countries, validating the role of data-driven approaches in energy planning. These findings underscore the importance of stable governance, targeted innovation support, and macroeconomic stability in promoting renewable energy production, providing policymakers in Europe with timely insights for achieving sustainable energy transitions. • Our paper uses econometrics and machine learning to analyze renewable energy production drivers and predict trends, supporting sustainable energy planning