A Comparison of forecasting model for export value of Thai’s cassava to China

Abstract

The purpose of the study is to compare the accuracy of the ARIMAX model (Autoregressive Moving Average with Exogenous Variables) with the BSTS models (Ba yesian Structural time series) and aims to forecast the export value of Thai's cassava to China. In the ARIMAX model, selecting an appropriate performance model found that ARIMAX (4, 2) (0, 0) was the most suitable with the lowest AIC and BIC value. The results of parameter estimation found that the exogenous variables including exchange rate and consumer price index had a statistically significant influence in the opposite direction on the value of Thai cassava exports product at the 95 percent confidence level. Whereas, the change in world oil price and world corn price cannot explain the value of Thai cassava exports to China. The results of the MAPE (Mean Absolute Percentage Error) findings of both are presented that the BSTS model performs suitable more than a classical model, the ARIMAX model with a comparison by a lower MAPE. The result for forecasting BSTS models predicted the value of Thai's cassava export to China for the next 12 months in the year 2021, the forecast model is volatile when compared to the actual value, which the model forecast a downward trend in the first half of the year and an upward trend in the last half of the year.