KAPSARC – A Machine Learning Framework for Global Energy Demand Prediction

This report presents a new econometric model using probabilistic machine learning to forecast global energy demand.

It covers approximately 150 countries using over 60 years of historical data and incorporates macroeconomic indicators, energy prices, and climate variables. The model demonstrates strong predictive accuracy and provides probabilistic forecasts with uncertainty quantification, allowing scenario analysis under uncertainty about future energy transition drivers.