<p>Recycled aggregate concrete, also known as RAC or RCA, is becoming more and more popular as a building material due to its favorable environmental characteristics. However, the use of RAC is becoming more and more hampered by the uncertainty surrounding its fracture resistance. 3 distinct estimation methods were considered in the present study, named multi-layered perceptron neural network (MLP), support vector regression (SVR), and adaptive neuro-fuzzy inference system (ANFIS) in order to appraise the splitting tensile strength (STS) of fiber-reinforced (Steel fiber, Carbon fiber, Polypropylene fiber, Basalt fiber, Glass fiber, and Woolen fiber) RAC. The dwarf mongoose optimization algorithm (DMOA) was linked with MLP, SVR, and ANFIS to the identification of the best-performing combination of hyperparameters. It was clear from sensitivity analysis that C, RORCA, and W have a significant impact on the prediction of STS at 0.9432, 0.9431, and 0.9201. The MLP_DM, SVR_DM, and ANF_DM algorithms provide significant promise for accurately predicting the STS of fiber reinforced RAC, as indicated by the findings. Throughout the training, validating, and testing phases, the ANFDM method showed outstanding dependability, with R² values of 0.9877, 0.9669, and 0.9818. The value of OBJ metric shows that the smallest value is for ANF_DM at 0.1028, then SVR_DM at 0.1578, followed by MLP_DM at 0.178.</p>