Development of a Forecasting Model for Thailand’s Frozen Shrimp Export Volume Using Machine Learning Techniques

Abstract

This research aims to develop a forecasting model for Thailand’s frozen shrimp export volume for January in 2026. The study applies machine learning techniques, namely Support Vector Machine Regressor (SVR) and Random Forest Regressor, to analyze the relationship between historical export volumes (measured in tons) and time variables in monthly format. The dataset used in this study is time-series data that has been preprocessed and refined to improve data quality prior to model development. The results indicate that the Random Forest Regressor model outperforms the SVR model in forecasting accuracy, achieving a coefficient of determination (R²) of 0.8984 and a lower error value (RMSE = 1622.27). In contrast, the SVR model demonstrates comparatively lower performance across all evaluation metrics. Furthermore, the selected model effectively captures trends and fluctuations in export volumes. The findings of this research can serve as a decision-support tool for strategic planning in production, supply chain management, and marketing strategies, ultimately enhancing the competitiveness of Thailand’s fisheries industry in the global market.