This paper presents the results of an investigation into several non-linear machine learning and soft
computing-based models, namely, feedforward neural network (FFNN), radial basis neural network
(RBNN), general regression neural network (GRNN), support vector machine (SVM), tree regression
fitting model (TREE) and adaptive neuro-fuzzy inference system (ANFIS). The data sets are from
extensive finite element modelling (FEM) of a shallow strip footing located near a homogeneous sandy
slope. The FEM outputs are used for training and testing the models. Furthermore, the predicted and
calculated models are compared and evaluated using different statistical indices and the most accurate
model is presented as a simple formula. After model evaluation process the most accurate model is
proposed to estimate the solution. The predicted results are compared with the FEM data, and a good
agreement is obtained representing good reliability for FFNN (R2=0.9233 for training and 0.9095 for
testing) solutions in this study. Moreover, the soft computing model is presented as a simple formula
and excellent agreement is obtained representing a high degree of reliability for the proposed model.