Assessment of compressive characteristics of coir fiber-reinforced concrete modified with wheat straw ash using predictive modeling

The growing demand for environmentally responsible construction materials requires reliable tools to predict the compressive strength of concrete incorporating agricultural waste and natural fibers. This study investigates the prediction of compressive strength of sustainable concrete incorporating wheat straw ash (WSA) and coir fiber (CF) using explainable ensemble machine learning. A database of 361 concrete mixtures was compiled from published experimental studies, considering cement, WSA, fine aggregate, coarse aggregate, water, superplasticizer, CF, and curing age as input variables. Random Forest, Gradient Boosting, LightGBM, and CatBoost models were developed and evaluated using an 80:20 training–testing split and five-fold cross-validation. All models demonstrated excellent predictive capability, with testing R 2 values exceeding 0.98. LightGBM achieved the best performance, with an R 2 of 0.9890, RMSE of 4.7541 MPa, and MAE of 3.0272 MPa. SHAP analysis identified superplasticizer, water, cement, fine aggregate, and curing age as influential parameters, while WSA and CF showed lower direct contributions. The proposed explainable framework supports efficient mixed optimization, reduces experimental effort, and promotes sustainable concrete development using agricultural waste materials