Predicting slump for high‐performance concrete using decision tree and support vector regression approaches coupled with phasor particle swarm optimization algorithm

Abstract The main focus of this study is to assess the slump characteristics of high‐performance concrete (HPC) using decision tree (DT) and support vector regression (SVR) models. In the first step, the models were solely fed via HPC samples to reproduce the slump rates. By coupling phasor particle swarm optimization (PPSO) to main models, hybrid DT‐PPSO and SVR‐PPSO frameworks, simulate the slump rates accurately. Using the correlation of determination and root mean square error (MAE) metrics for the DT, 96.04 and 5.097 were computed, respectively. SVR was obtained at 92.62 and 6.965, alternatively. In the hybrid approach, DT‐PPSO could improve by 3% and 55% in terms of correlation of determination and root MAE, respectively. DT‐PPSO appeared high‐accuracy model compared to others; however, a single DT had more desirable results than SVR. Overall, the advantages of this study encompass its methodological approach, comparative insights, and practical relevance, offering valuable contributions to the understanding and prediction of mechanical slump in HPC.