Modeling the Interaction Between Meteorological Indicators and Electricity Production Using Machine Learning Methods: The Case of Türkiye

Accurate forecasting of electricity production has become increasingly critical for energy systems, particularly in countries where renewable energy penetration and climate variability are rapidly increasing. In this context, this study investigates the interaction between meteorological indicators and electricity production in Türkiye using machine learning–based time series forecasting methods. The analysis is conducted using daily electricity production data obtained from the Turkish Energy Market Operator (EPİAŞ) and corresponding meteorological variables—namely apparent temperature, wind speed, and humidity—sourced from the Open-Meteo platform, covering the period from late 2015 to 2024. After aggregating hourly observations into daily series and applying appropriate preprocessing steps, several forecasting models are implemented and compared, including recurrent neural networks (RNN), long short-term memory (LSTM), gated recurrent unit (GRU), and the Prophet time series model. Model performance is evaluated using mean absolute error (MAE), root mean squared error (RMSE), mean absolute percentage error (MAPE), and the coefficient of determination (R2). The results indicate that gated RNN architectures significantly outperform classical approaches in modeling electricity production dynamics. Among all models, the GRU model achieves the best performance, yielding the lowest error values (MAE = 1241.4, RMSE = 1951.5, and MAPE = 3.5%) and the highest explanatory power (R2 = 76.9%). The LSTM model provides comparable but slightly weaker results, while the classical RNN and Prophet models exhibit substantially lower predictive accuracy, particularly in capturing nonlinear patterns and sudden production fluctuations. Overall, the findings demonstrate that GRU-based models offer a robust and reliable framework for electricity production forecasting in meteorologically sensitive energy systems and provide valuable insights for energy planning, grid operation, and data-driven energy policy development.