Enhancing Workability Prediction of Concrete Using Ensemble Machine Learning Models

This study integrates ensemble machine learning models with traditional concrete research methods to predict concrete workability. Leveraging a dataset of 103 laboratory experimental samples, key parameters such as Cement, Slag, and Water were explored for their impact on Slump. Random Forest, Gradient Boosting, and XGBoost Regressors demonstrated high accuracy, with XGBoost emerging as the most precise predictor$(\mathbf{R}^{\mathbf{2}}=\mathbf{0. 9 5 6}$). SHAP analysis revealed Water, Slag, and Coarse Aggregate as influential factors. This research pioneers an efficient and accurate methodology, showcasing the synergy of machine learning in civil engineering.