A Comparison of Time Series Forecasting Methods for electricity consumption of Uttaradit Rajabhat University

Abstract

The objective of this research was to compare of the time series forecasting for electricity consumption of Uttaradit Rajabhat University with 2 forecasting methods: Box-Jenkins method and Winters’ additive exponential smoothing method. The appropriate forecasting methods were chosen by considering the lowest of root mean square error. The time series data was divided into 2 sets. The first set had 108 months from January 2012 to December 2020 for constructing the forecasting models. The second set had 12 months from January to December 2021 for comparing accuracy of the forecasts. The results showed that the Winters’ additive exponential smoothing method was the better than the Box-Jenkins method in a short-term forecast for a period of 3 months but the Box-Jenkins was the better than the Winters’ additive exponential smoothing method in the middle-term forecast for a period of 6 months and the long-term forecast for a period of 12 months.