<![CDATA[management. However, comparative studies evaluating interpretable statistical models and ensemble learning algorithms for coffee shelf-life prediction remain limited, particularly using simulation-based datasets. This study compared the predictive performance of Multiple Linear Regression (MLR), Random Forest Regression (RFR), and Gradient Boosting Regression (GBR) using a simulation-based dataset of 400 observations representing realistic storage conditions. Model performance was evaluated using Mean Absolute Error (MAE), Root Mean Square Error (RMSE), and the coefficient of determination (R²). MLR achieved the best performance with the lowest MAE (10.01 days), the lowest RMSE (12.55 days), and the highest R² (0.8649), outperforming both ensemble learning models. Feature importance analysis consistently identified storage temperature as the most influential predictor of coffee shelf-life. These findings demonstrate that increased model complexity does not necessarily improve predictive accuracy and support the use of simulation-based datasets for developing predictive models prior to validation with experimental data.]]>
Copyrights © 2026
