TIME SERIES MODEL FOR FORECASTING THE QUANTITY OF THAI GLUTINOUS RICE FLOUR EXPORTED TO CHINA

Abstract

The objective of this research was to fit a forecasting model for the quantity of Thai glutinous rice flour exported to China. Statistical method used in this research was time series analysis and forecasting. The data used in this research were monthly data on the quantity of Thai glutinous rice flour exported to China from January 2016 through June 2022. The graph of time series data showed that these data had no seasonal pattern and their mean changed slowly. Therefore, the forecasting methods used in this research were the simple exponential smoothing technique and the nonseasonal Box-Jenkins method. Time series data were divided into two datasets. The first dataset was the first 72 observations which were used to fit the forecasting model. The results revealed that the MA(1) model derived from the Box-Jenkins method was the appropriate forecasting model. It had root mean square error less than the model derived from the simple exponential smoothing method. The second dataset was the last six observations used to evaluate the accuracy of the appropriate forecasting model using mean absolute percentage error. The result showed that the MA(1) model was the reasonable forecasting model.