Developing a boosted decision tree regression prediction model as a sustainable tool for compressive strength of environmentally friendly concrete

Abstract

One of the most significant parameters in concrete design is compressive strength. Time and money could be saved if the compressive strength of concrete is accurately measured. In this study, two machine learning models, namely, boosted decision tree regression (BDTR) and support vector machine (SVM), were developed to predict concrete compressive strength (CCS) using a complete dataset through the previous scientific studies. Eight concrete mixture parameters were used as the input dataset. Four statistical indices, namely the coefficient of determination (R²) and root mean square error (RMSE), mean absolute error (MAE), and RMSE-Standard Deviation Ratio (RSR), were used to illustrate the efficiency of the proposed models. The results show that the BDTR model outperformed SVM model with the overall result of R²=0.86 and RMSE=6.19 and MAE=4.91 and RSR=0.37, respectively. The results of this study suggest that the compressive strength of high-performance concrete (HPC) can be accurately calculated using the proposed BDTR model.

Keywords: Boosted decision tree regression (BDTR); Concrete strength prediction; Machine learning model.