Table 1 Comparative summary of recent ML-Based biodiesel studies.
Author’s | ML algorithms used | Dataset description | Validation method | Performance metrics | Key findings |
|---|---|---|---|---|---|
Sanjeevannavar et al.42 | XGBoost, Random forest | Engine performance data from biodiesel blends | Train-test split | R2 = 0.99, RMSE = 0.54, MSE = 0.24 MAE = 0.29 | XGBoost outperformed; emphasized hyperparameter tuning |
Siqueira-Filho et al.43 | XGBoost | Fuel consumption data from thermoelectric plants | Train-test split | R2 = 0.95 | XGBoost showed high accuracy and computational efficiency |
Navid Kardani et al.44 | XGBoost, ANN and SVM | conversion of lignocellulosic biomass during hydrothermal carbonization | Train-test split | R2 = 0.999 | XGBoost showed high accuracy |
Xiangmeng Chen et al.45 | ANN, GAM, SVR and ELM | Conversion of bio-oil from biomass by catalytic pyrolysis | Train-test split | R2 = 0.89, RMSE = 0.03, and MAE = 0.01 | ELM showed high accuracy |
Xin Jin et al.46 | kNN, SVM, RF and AdBoost | Biodiesel yield prediction | Train-test split | Training RMSE = 2.778 Validation RMSE = 5.178 | RF showed high accuracy |
Ilham Yahya et al.47 | LR, RF, ANN and SEM (Stacking ensemble models) | Microwave-assisted biodiesel yield prediction | Train-test split | R2 = 0.949 and lowest MSE of 54.64. | SEM showed high accuracy |
Almohana et al.48 | AdBoost-HBR, AdBoost-DT, AdBoost-GBR, | Biodiesel conversion from WCO | Train-test split | R2 = 0.996, MAE = 1.82 | AdBoost-GBR showed high accuracy |
Abdulrahman Sumayli49 | Boosted-MLP, GPR, KNN | Biofuel production from papaya oil | Train-test split | R2 = 0.994 RMSE = 6.5180 | Boosted KNN showed high accuracy |
Somboon Sukpancharoen et al.50 | SVR, RF, XGB, and KRR | Biodiesel production | 10-fold cross-validation | R2 = 0.98 | XGB showed high accuracy |
