Forecasting Export Values of Cars, Equipment and Parts of Thailand by Time Series Methods

Abstract

This research aimed to study the forecasting methods for the export values of cars, equipment and parts of Thailand by time series methods including moving average, Holt-Winters exponential smoothing and SARIMA. The monthly export value data were gathered from Office of the Permanent Secretary Ministry of Commerce from January 2008 to December 2019, a total of 144 months. The data were divided into two sets. The first set was the data from January 2008 to December 2018 and was used to develop the forecasting models. The second set was the data from January 2019 to December 2019 and was used to compare the forecast accuracy by the mean absolute percentage error (MAPE). The results of study showed that the SARIMA method had MAPE at 10.18% while the Holt-Winters exponential smoothing and the moving average methods that had MAPEs at 10.38% and 11.06%, respectively. Thus, the SARIMA method was the most appropriate and had the highest accuracy on forecast of the export values of cars, equipment and parts of Thailand.