Problems in deep excavations are full of uncertain, vague, and incomplete information. In most instances, successfully solving such problems depends on experts' knowledge and experience. The primary object of this research was to propose an "Evolutionary Support Vector Machine Inference Model (ESIM)" to predict wall deformation in deep excavation in Taipei Basin. ESIM is developed based on a hybrid approach that fuses support vector machines (SVM) and fast messy genetic algorithm (fmGA). SVM is primarily concerned with learning and curve fitting; and fmGA with optimization. Fifty-seven wall deformation monitoring database were collected based on monitoring data and compiled from prior projects. Fifty-two of 57 were selected for training, leaving 5 valid cases available for testing. Results show that ESIM can successfully predict the deflection and apply to contractors utilizes the knowledge and experience from past projects to predict wall deformation of new projects. Therefore the construction and foundation construction contractors can update wall deflection monitoring data of different stages during deep excavation process, in order to predict the wall deformation of the next stage and examine whether the max deflection is within the controlled range. The results are used as guidelines on site safety and risk management.