<![CDATA[Nile tilapia aquaculture is one of the aquaculture subsectors with significant development potential. However, the productivity of Nile tilapia cultured in brackishwater ponds is often constrained by variability in technical factors such as the number of fingerlings stocked, pond area, stocking density, land status, planting season, and feed quantity. To address these challenges, a predictive model based on machine learning was developed. Data were collected through field observations and interviews with Nile tilapia farmers in Wanantara, Sindang, Indramayu. The data were then processed using label encoding and normalization techniques. The dataset was divided into 80% for training and 20% for testing. XGBoost, Random Forest, and Support Vector Regression algorithms were trained using hyperparameter tuning and five-fold cross-validation, and evaluated using RMSE and R² metrics. The results show that XGBoost achieved the best performance (R² = 0.9798 and RMSE = 442.05 kg), followed by Random Forest (R² = 0.955 and RMSE = 679.742 kg) and SVR (R² = 0.888 and RMSE = 1065.367 kg).]]>
Copyrights © 2026
