Fig 1.
Methodology of the study.
Table 1.
Hyperparameter tuning details and final configurations.
Table 2.
k-Fold cross-validation setup and implementation.
Fig 2.
Correlation heatmaps between actual and predicted 28-day compressive, tensile, and flexural strengths for ML models.
Fig 3.
Model performance heat map for all models.
Fig 4.
Correlation matrix.
Fig 5.
Scatter plots for predicted values of train and test of CS, TS, and FS.
Table 3.
Descriptive analysis.
Fig 6.
Plots of feature importance for CS, TS, and FS.
Fig 7.
Box plots of CS, TS, FS vs biomedical waste ash (%).
Fig 8.
Radar chart for train and test performance metrics.
Fig 9.
3D surface plots for CS, TS, and FS, respectively.
Fig 10.
Error Analysis for all the models.
Fig 11.
Sensitivity curve for CS, TS, and FS.
Fig 12.
Shap analysis.
Fig 13.
Shap value plot.
Fig 14.
Taylor analysis for train and test values for all models.
Fig 15.
Pareto front plot for CS, FS, and TS, respectively.
Fig 16.
Pairwise plot for CS, FS, and TS, respectively.
Table 4.
External validation results.
Table 5.
SI for BMWA concrete mixes.