Enhancing Forecast Accuracy of Thailand’s Seasoning Exports Using SARIMA Models with Grid Search and Bootstrap Prediction Intervals

Accurate forecasting of Thailand’s seasoning exports to ASEAN+8 and global markets plays a vital role in developing effective trade strategies and managing economic risks. This research compares SARIMA models for export forecasting, utilizing monthly export data from 2013 to 2024. Specifically, we evaluate SARIMA model selection techniques-Grid Search (based on AIC and BIC) versus automatic SARIMA selection using the auto.arima() function in R. The results demonstrate that SARIMA models identified through Grid Search deliver superior accuracy, especially in short-term forecasting, where errors are reduced by approximately 30%. Additionally, the use of Bootstrap Prediction Intervals outperforms Standard Prediction Intervals, offering more flexible and realistic measures of uncertainty that are well-suited to volatile market environments. By integrating optimal SARIMA modeling with Bootstrap Prediction Intervals, policymakers and industry stakeholders gain enhanced reliability for strategic planning and risk management decisions in international trade. This approach significantly strengthens decision-making capabilities in an uncertain economic context.