Abstract

This study examines various predictive models for forecasting electricity demand in the ASEAN region, characterized by rapid industrialization and diverse economic landscapes. Traditional methods like ARIMA and the Grey Model, while effective in certain scenarios, often fall short in capturing the complexities of electricity consumption due to evolving factors like technological advancements and economic policies. Advanced models such as Neural Network Autoregressive, Echo State Networks, Autoregressive Support Vector Machine, and Extreme Gradient Boosting have demonstrated superior capabilities in various fields, including finance and public health. Our research, leveraging data from the World Development Indicator database, evaluates these models specifically for per capita electricity consumption in ASEAN nations. The results highlight the advanced models, particularly XGBoost, for their accuracy and adaptability in handling the dynamic electricity demand, suggesting their greater efficacy over traditional models. This study not only provides valuable insights for electricity demand forecasting in rapidly evolving economies, but also emphasizes the importance of appropriate model selection for informed energy policy and planning. The findings have significant policy implications, aiding in the development of more efficient, sustainable, and equitable energy management across the ASEAN region, and pave the way for future research integrating real-time data and socio-economic factors.

Authors

Samuel John E. Parreno
University of Mindanao Digos College, Philippines

Keywords

Electricity Demand Forecasting, Predictive Modeling, ASEAN Region, Machine Learning, Time Series Analysis