فهرست نشریات علمی دسترسی آزاد ایران

This study presents a data-driven framework for predicting the compressive strength of fly ash concrete using the Extreme Gradient Boosting (XGBoost) algorithm. A dataset of 1,030 experimental samples was used for model development, and an additional 198 independent samples from external studies were employed for validation. Seven key input parameters—cement, fly ash, water, superplasticizer, coarse aggregate, fine aggregate, and curing age—were considered as predictors. To ensure robustness, a leakage-free preprocessing strategy and randomized search (RandomizedSearchCV) for hyperparameter tuning were applied for hyperparameter tuning. The optimized model achieved excellent predictive accuracy with R^2=0.9949 and RMSE = 1.20 MPa on the training set, and R^2=0.9390 with RMSE = 3.96 MPa on the testing set. When evaluated on the independent validation dataset, the model maintained acceptable accuracy (R^2=0.6745; 73.7% within ±20% tolerance). Feature importance analysis identified curing age and cement content as dominant factors influencing compressive strength. The results demonstrate that while machine learning models can achieve near-perfect accuracy internally, external validation is essential for assessing real-world generalization. The proposed XGBoost framework offers a reliable, interpretable, and efficient tool for data-driven mix design and sustainable concrete optimization.