Strength Prediction Models for Concrete Incorporating Fly Ash as a Waste Material

A comparative machine learning–based methodology was adopted to predict the compressive strength of fly ash concrete using Multiple Linear Regression (MLR), Support Vector Regression (SVR), AdaBoost Regressor (ABR), Random Forest (RF), and Extreme Gradient Boosting models (XG). A dataset of 498 mix designs collected from published literature was used, considering cement, fine and coarse aggregate, fly ash content, water content, water–cement ratio, and curing period as input parameters. Model performance was evaluated using mean absolute error, root mean square error, and coefficient of determination. The Extreme Gradient Boosting model showed the best predictive capability (R² = 0.881; RMSE = 5.65 MPa). Sensitivity analysis identified curing period, cement content, and water content as the most influential variables. The results demonstrate reliable strength prediction and enable model comparison to support data-driven mix optimization for sustainable fly ash concrete (FAC).