<![CDATA[Purpose: The objective of this research is to provide more accurate predictions and transparency in the analysis process, making it easier for engineering professionals to comprehensively interpret, validate, and evaluate the model results. Methods: The algorithm used is XGBoost with the XAI-SHAP approach, which serves to interpret the contribution of each input feature, like the water-cement ratio, cement content, and concrete age, to the predicted output. This approach ensures that the model is not only accurate but also open and accountable to engineering practitioners. The research methods are data collection, model training, SHAP visualization, and performance evaluation using RMSE, MAE, and R² metrics. The dataset used is the Concrete Compressive Strength dataset from the UCI Repository, which consists of 1,030 records. Result: Subsequent to model training using the XGBoost with optimal hyperparameters identified via Random Search, model performance was evaluated across three scenarios: training data, testing data, and 5-fold cross-validation. The results indicate that the proposed XGBoost exhibits show high performance with an RMSE 4.59 ± 0.63 and R² of 93% as well as stability, which is evidenced by consistent performance metrics, without notable signs of overfitting. Novelty: The novelty of this study lies in the combination of XGBoost with the XAI-SHAP approach for predicting concrete compressive strength, providing both accuracy and interpretability that can work with nonlinear data, as well as its ability to produce results that are easily understood by construction professionals, making it a favorable choice for predicting concrete compressive strength.]]>
Copyrights © 2026
