Abstract
Annual fluctuations in rice production present significant challenges for agricultural planning and regional food security, particularly in areas where agricultural productivity is strongly influenced by environmental and production-related factors. Developing an accurate forecasting model is therefore essential to support evidence-based decision-making and improve agricultural resource management. Although Support Vector Regression (SVR) has demonstrated promising performance in agricultural prediction, previous studies have primarily focused on large-scale datasets or relied on default model parameters, limiting their applicability to local agricultural conditions. Therefore, this study aims to develop and evaluate a GridSearchCV-optimized Support Vector Regression model for predicting annual rice production in Manokwari Regency. Secondary data were obtained from the Agricultural Statistics Database maintained by the Indonesian Ministry of Agriculture, rainfall records from Rendani Meteorological Station, and statistical publications issued by the Central Statistics Agency of Manokwari Regency. The proposed framework integrates data preprocessing, systematic hyperparameter optimization using GridSearchCV, and Support Vector Regression modelling. The predictive capability of the model was assessed using Mean Absolute Error (MAE), Root Mean Squared Error (RMSE), Mean Absolute Percentage Error (MAPE), and the coefficient of determination (R²). The results demonstrate that the selected model employed the Radial Basis Function (RBF) kernel with C = 1000, gamma = 0.01, and epsilon = 0.001, achieving an MAE of 460.15, RMSE of 718.86, MAPE of 5.36%, and an R² value of 0.9578. These findings demonstrate satisfactory predictive performance and suggest that the proposed model has the potential to serve as a decision-support tool for agricultural production planning and regional food security management in Manokwari Regency.