A data mining approach for proposing a relationship to predict self-compaction concrete crack width after the self-healing period

In this study, three powerful soft computing methods, including Group Method of Data Handling (GMDH), Response Surface Methodology (RSM), and Gene Expression Programming (GEP), were used to evaluate the effects of fly ash (FA), limestone powder (LP), silica fume (SF), and initial crack width (CWB) on the crack width after the self-healing period (CWA) and to propose a relationship for its prediction. The results showed that all three models had a good ability to provide a relationship for predicting CWA. The GMDH model reduced the RMSE, MAE, and StD by 24.8 %, 24.3 %, and 18.6 %, respectively, compared to the RSM model. Also, compared to the GEP model, the reductions for RMSE, MAE, and StD were 23.9 %, 12.7 %, and 62.8 %, respectively, indicating its superior performance in predicting CWA. This model achieved the lowest errors, with MSE, RMSE, and MAE values of 48.5713, 6.969, and 4.878, respectively, indicating its high accuracy in prediction and error minimization. Sensitivity analysis displayed that the CWB had the most significant impact on the model output, CWA. In descending order of importance, the most influential parameters affecting CWA were LP, SF, and FA.