Table 6 Comparative summary of machine learning models for predicting mechanical and durability properties of sustainable concretes in previous literature and the present study.
Reference study | Material system | Algorithm | Target property | R2 (best model) | Remarks |
|---|---|---|---|---|---|
Fly-ash blended concrete | Random Forest | Compressive strength | 0.93 | Ensemble models generalize well on small data | |
GGBS-silica fume concrete | XGBoost | Compressive strength | 0.96 | Boosted trees capture nonlinear SCM effects | |
Fiber-reinforced concrete | XGBoost | Chloride permeability | 0.99 | Outperforms deep CNN for durability indices | |
Natural-fiber concrete | CNN–LSTM | Tensile strength | 0.60 | Sequence models less effective for static inputs | |
Present study (2025) | FA–GGBS–Coir–GNP concrete | XGBoost | All targets (Compressive Strength 28 days, Split tensile strength 28 days and RCPT) | 0.74–1.00 | Superior balance of accuracy and interpretability |
