Title: Machine learning approach to forecasting global energy consumption and sustainability

Authors: George Sammour; Anas Irsheid

Addresses: Princess Sumaya University for Technology, Khalil Al-Saket Street, Al-Jubaiha, P.O. Box 1438, Amman 11941, Jordan ' Umniah Mobile Company, Queen Nour St., Shmesani, P.O. Box 942481, Amman 11194, Jordan

Abstract: This study applies machine learning specifically K-means clustering and long short-term memory (LSTM) networks to analyse and forecast global energy consumption patterns. Countries were classified into 12 clusters based on energy profiles, GDP, and environmental impacts, identifying seven unique national profiles (the USA, China, France, Germany, Russia, Japan, Canada, and Brazil). These clusters revealed varying dependencies on fossil fuels and renewable sources. LSTM models predicted primary and renewable energy consumption, offering insights for energy planning and investment. Results indicated distinct reliance patterns, renewable energy forecasts were more accurate in Germany and Canada due to stable infrastructure and policies, while Russia and Brazil showed higher variability in renewable adoption. Visual analyses captured deviations during the 2008-2009 financial crisis and the 2020 COVID pandemic. Overall, the findings demonstrate machine learning potential to optimise energy management, support the United Nations Sustainable Development Goals, and mitigate climate change through improved forecasting and resource utilisation.

Keywords: energy consumption forecasting; machine learning in energy; fossil fuel dependency; sustainable energy strategies; UN-SDGs; energy demand prediction; long short-term memory; LSTM; K-means clustering.

DOI: 10.1504/IJCEE.2026.155824

International Journal of Computational Economics and Econometrics, 2026 Vol.16 No.3, pp.218 - 241

Received: 05 Apr 2025
Accepted: 27 Aug 2025
Published online: 24 Aug 2026 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article