Ski-jump spillway (SJS) is one of the energy-dissipating structures, passing the flood through the dams. In these structures, the downstream bed is scoured due to the collision of the high energy jet. It is essential to obtain information on the scour-hole dimensions to enhance the safety of the dam and related structures. The machine learning and data mining methods are efficient in predicting hydraulic phenomena. In this study, the geometric properties of the scour-hole including the hole expansion (width), maximum depth of the scour and its distance from the end edge of the spillway were predicted using a model by combining the adaptive neuro-fuzzy inference system (ANFIS) and the particle swarm optimization (PSO) algorithm. The results showed that the developed ANFIS–PSO model obtained maximum scour depth with a precision of R2 = 0.94 and RMSE = 0.565, scour hole width with precision of R2 = 0.95 and RMSE = 1.153 and its distance from the end edge of the SJS with an accuracy of R2 = 0.96 and RMSE = 1.809. The evaluation and comparison of the simulation results indicated the superiority of the proposed model over other methods.