Predicting the compressive strength of concrete from its compositions and age using the extreme gradient boosting method

Abstract

The uniaxial compressive strength (UCS) is one of the most important mechanical properties of concrete. This paper aims to demonstrate that the UCS of concrete can be accurately predicted from its compositions and age using the extreme gradient boosting regression (XGB) method. The artificial neural networks (ANN) and the support vector machine (SVM) methods are also considered to compare with the XGB method. A relevant laboratory measurement dataset available in literature is considered to train and test the machine learning (ML) methods. We observe that all the three considered ML methods provide accurate results. However, the XGB method is more robust, faster to train and more accurate than the ANN and SVM methods as well as other existent ML methods presented in literature.