Forecasting energy consumption in Mozambique: A comparative analysis of advanced machine learning models from 2025 to 2045

This research aims to provide a robust foundation for future energy infrastructure development and sustainability efforts in Mozambique. Accurately forecasting energy consumption is crucial for the strategic planning and sustainable development of energy infrastructure, particularly in emerging economies like Mozambique. This study employs advanced machine learning models—XGBoost, Neural Networks, Gradient Boosting Regression, Elastic Net, and Random Forest—to predict Mozambique’s energy consumption from 2025 to 2045. By comparing the predictive accuracy of these models using error metrics such as Mean Absolute Error (MAE), Mean Squared Error (MSE), and Root Mean Squared Error (RMSE), the research identifies the most effective tools for future energy planning. The results highlight the superiority of the Random Forest model, which consistently achieved the lowest error rates, suggesting it as the most reliable model for capturing the complexities of energy demand in Mozambique. In contrast, models like XGBoost demonstrated higher error rates, indicating potential limitations in their application to this dataset. The findings of this study provide valuable insights for policymakers and industry stakeholders, contributing to the development of more accurate and reliable energy forecasts, which are essential for ensuring the sustainable growth of Mozambique’s energy sector.