Support vector machine-based prediction model for the compressive strength for concrete reinforced with waste plastic and fly ash

In the current study, the effect of the inclusion of waste plastic in different quantities (0–10%) on the compressive strength of fly ash reinforced concrete is explored. Compressive strength decreases with increasing plastic waste due to its inert, hydrophobic nature and poor bonding within the concrete matrix, while 10% fly ash improves strength slightly through pozzolanic reactions that densify the matrix and reduce voids. Support vector machine (SVM) was explored as a potential machine learning technique for accurately predicting and modeling the compressive strength of concrete. Predictive models were developed using SVM-radial basis function (RBF), SVM-linear, SVM-power and linear regression. The models were analyzed using performance metrics such as mean absolute error (MAE), mean squared error (MSE), root mean squared error (RMSE), mean squared logarithmic error (MSLE), root mean squared logarithmic error (RMSLE), coefficient of determination (R2), mean absolute percentage error (MAPE), Willmott's index of agreement, Mielke and Berry index, and Legates and McCabe's index. Taylors diagram was also used to analyze the models. SVM-RBF model outperformed all other models with an R2 value of 0.969 and 0.771 in training and testing respectively. Sensitivity analysis revealed that % plastic waste is the most influential parameter in predicting the compressive strength of concrete with a score of 59.04%.