| dc.description.abstract | Accurate prediction of rice crop yield is crucial for supporting food security, especially in rice-producing regions like Serdang Bedagai Regency, North Sumatra. Rice crop yields are influenced by various factors, including climatic conditions (air temperature, rainfall, relative humidity, sunshine duration, and wind speed), the size of the cultivated area, and the yields from previous periods, which provide historical productivity patterns. This research aims to develop a rice crop yield prediction model by combining the Random Forest (RF) algorithm with the Whale Optimization Algorithm (WOA) to enhance prediction accuracy. The data utilized includes weather data (temperature, rainfall, humidity, etc.) and historical rice crop yield data from 2012 to 2022. The research stages encompass data pre-processing, feature selection, model training, and parameter optimization using WOA. The results indicate that the WOA-RF model achieved the best performance with an R-squared (R²) value of 0.9945, a Mean Absolute Error (MAE) of 356.17 tons, and a Mean Absolute Percentage Error (MAPE) of 1.08%. This model outperformed standard Random Forest, XGBoost, and Support Vector Regression (SVR) in terms of accuracy and consistency. These findings demonstrate that parameter optimization with WOA significantly improves the predictive capability of the RF model. The developed web-based prediction system can serve as a valuable tool for farmers and policymakers in more effective agricultural planning. | en_US |
