Cellulose from the oil palm empty fruit bunch: Optimization of extraction using response surface methodology and support vector machine

Oil Palm Empty Fruit Bunch (OPEFB), a major lignocellulosic by-product of the palm oil industry, contains high amounts of cellulose and is widely recognized as a promising raw material for sustainable biomass utilization. The optimization of OPEFB was explored using Response Surface Methodology (RSM) and Support Vector Machine (SVM) algorithms. The predictive performance of both models was evaluated using statistical metrics, including R², MAE, MSE, and RMSE. The RSM demonstrated superior prediction accuracy, as evidenced by lower MSE (0.027), RMSE (0.165), and MAE (0.140) values, compared to the SVM model (MSE = 0.152, RMSE = 0.390, and MAE = 0.318). In addition, the coefficient of determination (R²) for RSM was higher (0.998) than that for SVM (0.990), indicating that RSM provided a better goodness-of-fit and explained a greater proportion of variability in the cellulose yield. The larger error values and lower R² observed in the SVM suggest that the model produced greater deviations from the experimental results, despite still showing acceptable predictive capability. These findings confirm that RSM provided more stable, consistent, and accurate predictions of cellulose yield under the experimental conditions applied in this study. The optimal conditions for achieving the optimized cellulose yield were 19.73% (w/v) NaOH, 9.52% (w/v) H₂O₂, 22.14% (w/v) KOH, and 69.86% (w/v) CH₃COOH, achieving cellulose contents of 89.74% (RSM) and 89.69% (SVM). Morphological characterization using SEM revealed the transformation of the cellulose structure from compact and dense to fibrous and porous with increasing chemical treatment intensity. FTIR analysis confirmed the removal of lignin and hemicellulose, while XRD indicated an increase in crystallinity from 19.48% to 64.95% and a shift towards the cellulose structure. This study highlights the potential of integrating RSM and SVM to optimize cellulose extraction, offering insights into the complex interactions among process variables and their effects on cellulose yield and purity.