Extreme gradient boosting model for forecasting slump and compressive strength of highperformance concrete

Abstract

The Extreme Gradient Boosting (XGBoost) model has found success across a wide range of engineering challenges because it is straightforward, flexible, and applicable to both classification and regression tasks. Concrete, a composite material made up of several intricate components, is affected by various factors, making it difficult to predict its properties with precision. This article presents an Extreme Gradient Boosting (XGBoost) model designed to predict the slump and strength of high-performance concrete (HPC) incorporating a combination of blast furnace slag and silica fume as mineral admixtures. The model was developed using an experimental dataset following the Box-Hunter statistical method. The criteria used to assess the models’ accuracy are R squared (R_²) and mean squared error (MSE). The findings indicate that the XGBoost model, developed using an experimental dataset following the Box-Hunter statistical method, is well-suited for predicting both the slump and compressive strength of concrete.