Optimization remains one of the most critical issues in civil engineering studies. In this case, the study focused on the development of three types of Artificial Neural Networks (ANNs), which included the Multilayer Perceptron (MLP), Neuro-Fuzzy (NF), and Radial Basis Function (RBF). The four input parameters for each model have been used while the output parameter remained the predicted final profit. For the development of MLP models, the use of one or two hidden layers with varying number of neurons was considered to identify the best structure of MLP network. The most efficient NF network was obtained by modeling using various membership functions, depending on error indices. Because ANNs function in a closed loop and the impact of input parameters on output cannot be indicated in any form, an uncertainty analysis was performed to find out the absolute derivatives of the output parameter with regard to the four input parameters. Following this, the relative derivatives of the output regarding each of the input parameters were analyzed. It was found that the MLP neural network was the best performing model among all of the models studied. In addition, the sensitivity analysis found that pre-sale financing method was the most significant input parameter.