Hybrid machine learning model and Shapley additive explanations for compressive strength of sustainable concrete

The application of the traditional support vector regression (SVR) model to predict the compressive strength of concrete faces the challenge of parameter tuning. To this end, a hybrid machine learning model combines the SVR model and grid search (GS) optimization algorithm, namely the GS-SVR model was proposed and employed on 559 datasets with eight input variables to achieve the compressive strength prediction of sustainable concrete. Moreover, the prediction performance was compared with the original SVR model. Results showed that the GSSVR model outperformed the original SVR with R-2 = 0.93, R-2 = 0.85, MAE = 3.52, MAE = 5.22, and RMSE = 4.89, RMSE = 7.01, respectively. The model proposed can be recommended as a reliable and accurate compressive strength prediction tool to assist or partially replace laboratory compression tests to save cost and time. Additionally, the Shapley additive explanation (SHAP) method was used to explain the importance and contribution of the input variables that influence the compressive strength.

Automated summary

A hybrid machine learning model called GS-SVR, combining the SVR model and grid search optimization algorithm, was proposed to predict the compressive strength of sustainable concrete. The results showed that the GS-SVR model outperformed the original SVR model and can be recommended as a reliable and accurate compressive strength prediction tool. Additionally, the SHAP method was used to explain the importance and contribution of the input variables that influence the compressive strength.