Short-Term Demand Forecasting for Energy in Thailand

Abstract

The purpose of this research was comparing of the most appropriate forecasting methods for predicting of short-term energy demand in Thailand. The data, gathered from Energy Policy and Planning Office, Ministry of Energy between July 2017 and September 2020, of 39 values were used and separated into two groups. The first group was contained 30 values between July 2017 and December 2019 for studying and comparing the most appropriate forecasting methods by (1) Moving Average Method, 2) Trend Analysis Method, 3) Single Exponential Smoothing Method, 4) Double Exponential Smoothing Method, 5) Triple Exponential Smoothing Method, and 6) Decomposition Method. The suitable forecasting method was chosen by considering the smallest value of Mean Absolute Percent Error (MAPE) and Mean Absolute Deviation (MAD). Then the selected most appropriate method was used to determine the most suitable forecasting period by the second group which was contained 9 values from January 2020 to September 2020 for finding the most appropriate for the predictive timing. The result indicated that Decomposition Method was the best and suitable for forecasting in advance of 3 months.