A holistic methodology for predictive modeling of total final energy use in Türkiye till 2030

Abstract

This study introduced novel per capita-based regression models to forecast T & uuml;rkiye's annual total final energy use across the aggregated transportation, industrial, and service sectors, households, and all sectors up to 2030. Model accuracy was validated against historical data (2018-2022) and the 2022 National Energy Plan (NEP) targets, demonstrating strong short- and mid-term performance with Mean Absolute Percentage Errors (MAPEs) of 2.8 %-5.6 %. Forecasts for 2025 (5400-6070 PJ-PJ) showed reasonable agreement with the NEP (5190 PJ), while longer-term (2030) projections (5530-6760 PJ) exhibited increasing divergence from the NEP (5840 PJ) and a prior study. The per capita approach demonstrated robustness through significant agreement with advanced (exponential smoothing) and basic (linear) classical forecasting methods, highlighting the substantial influence of population dynamics on energy demand. Acknowledging the limitations of a short historical dataset and the assumption of stable per capita consumption, this study provides a valuable and validated tool for short to medium-term energy planning in T & uuml;rkiye. The findings offer a crucial baseline for future research integrating broader socioeconomic indicators, sectoral disaggregation, and the impact of specific policy and technological factors to enhance long-term forecasting accuracy and inform T & uuml;rkiye's evolving energy roadmap.