Forecasting the Prices of Jasmine 105 Paddy Rice

Abstract

The objective of this research was to forecast the prices of Jasmine 105 paddy rice using time series data from the website of Office of Agricultural Economics with total 212 values during January, 1997 to August, 2014. We divided this time series data into 2 sets, the first set had 204 values from January, 1997 to December, 2013 for constructing the forecasting models by 4 time series analysis techniques: Box-Jenkins method, damped trend exponential smoothing method, combined forecasting method using weights based upon the ordinary least squares regression coefficients, and combined forecasting method using weights based upon the proportion of the values in the eigenvector from the principal component analysis. The second set had 8 values from January to August, 2014 for comparing accuracy of the forecasts via the criteria of the lowest mean absolute percentage error and root of mean squared error. Research findings indicated that for all forecasting methods that had been studied, combined forecasting method using weights based upon the ordinary least squares regression coefficients was the most appropriate method for this time series.