The Application of Time Series Models Considering Seasonality in Monthly Electricity Production Forecasting

This study comprehensively examines the trends in global electricity demand and assesses the efficacy of various forecasting models applied to monthly electricity usage data from Vietnam, Thailand, South Africa, and Turkey between 2010 and 2023. Four predominant time series forecasting models—Holt-Winters, Seasonal Autoregressive Integrated Moving Average (SARIMA), FB-Prophet, and Neural Basis Expansion Analysis for Time Series (N-BEATS)—are evaluated, providing a detailed comparative analysis based on their performance in datasets characterized by diverse seasonal and non-seasonal patterns. The effectiveness of each model is quantified using two error metrics: Mean Absolute Percentage Error (MAPE) and Root Mean Square Error (RMSE). The findings indicate that the FB-Prophet model is particularly effective in data environments with stable seasonal trends, such as those observed in Vietnam and Thailand, whereas the N-BEATS model shows superior performance in handling complex data patterns, which is evident in South Africa and Turkey. This underscores the critical importance of model selection based on specific data characteristics and highlights the potential of integrating statistical and deep learning approaches to improve forecasting accuracy. The implications are significant, offering strategic insights for operational planning and decision-making in energy markets, especially amid the increasing integration of renewable energy sources and the changing dynamics of global electricity demand.