Abstract <p>Over the past 20 years, the value of Thailand’s fresh durian exports has steadily increased, establishing durian as a significant agricultural export product. This study aims to identify the key factors influencing Thailand’s durian export value and develop accurate forecasting models using a combination of statistical and machine learning methods. Quarterly data from 2002 to 2022, spanning 84 quarters, was analyzed using four statistical approaches—Winter’s Exponential Smoothing, Seasonal Autoregressive Integrated Moving Average (SARIMA), Multiple Linear Regression, and Seasonal Autoregressive Integrated Moving Average with Exogenous Variables (SARIMAX)—and two machine learning techniques—Support Vector Regression (SVR) and Multilayer Perceptron Neural Networks (MLP). A novel combined forecasting method, hybrid models, integrating statistical and machine learning models, was also developed to enhance predictive accuracy. The results revealed that the volume of Thailand’s durian exports and China’s gross domestic product (GDP) are significant factors positively correlated with the export value. In terms of forecasting accuracy, machine learning methods demonstrated superior performance compared to traditional statistical models. Notably, the combined SARIMAX-MLP model delivered the highest accuracy, outperforming both individual approaches. This study offers an innovative contribution by integrating statistical and machine learning techniques to improve forecasting precision for Thailand’s durian export value. These findings are valuable for policymakers, businesses, and researchers involved in agricultural export planning and management. The study also provides a foundation for future research, with potential applications of these methods to other agricultural products and sectors.</p>