Pile settlement (SP) socketed to rock has received substantial attention. Despite introducing some design methods of SP prediction, considering the novel and practical predicting model with satisfactory performance is pivotal. The primary target of the present paper is to find out the applicability of applying two hybrid adaptive neuro-fuzzy inference system (ANFIS) models in predicting the pile settlement (SP) socketed to rock in the Klang Valley Mass Rapid Transit (KVMRT) project constructed in Kuala Lumpur, Malaysia. Two constructive variables of the ANFIS method are tuned using two optimization algorithm named antlion optimization (ALO) and grasshopper optimization algorithm (GOA). Five parameters, including the ratio of length in soil layer to length in the rock layer, total length to diameter, uniaxial compressive strength, standard penetration test, and ultimate bearing capacity, were considered as input variables. The results specify remarkable potential in both the training and testing phases for both hybrid ANFIS models. Comparing the measured settlement with those predicted by ALO-ANFIS and GOA-ANFIS methods demonstrated that the developed models have R-2 larger than 0.9077 for training data and bigger than 0.9387 in the testing phase. ALO-ANFIS has better performance than GOA-ANFIS in all four statistical performance evaluators. Both of the hybrid ANFIS models have a high capability for predicting SP. However, ALO-ANFIS represents the more prominent ability to determine the optimal value of the ANFIS parameters than other ones.
