PREDICTION OF BOESENBERGIA ROTUNDA (L.) DRYING USING HOT AIR AND ULTRASONIC VIBRATION VIA EXTREME GRADIENT BOOSTING

This study used eXtreme Gradient Boosting (XGBoost) to predict the drying behavior of Boesenbergia rotunda (L.) under hot air and ultrasonic vibration. A dataset of 73 samples with temperature, time, and vibration as inputs was used. The model achieved high accuracy with R² = 0.99 and RMSE = 8.71 on test data. SHAP analysis revealed that ultrasonic vibration improved moisture reduction, especially during longer drying times. These results demonstrate the effectiveness of combining XGBoost and SHAP for understanding and optimizing complex drying processes in food and agricultural engineering.