Comparison of the Bias andWeighting of Variables in Neural Networks (ANN) for the Selection of the Type of Housing in Spain and Mexico

Abstract

This chapter compares users’ housing characteristics in Spain and Mexico through a multilayer neural network trained for selecting the right type of housing by new users. This research aims to analyze the biases and synaptic weights of the variables that are analyzed. Our results showthat data’s bias and variables’weighting do not influence the neural network’s precision for housing classification. Thus, the housing classification is independent of the biases and captures the housing users’ preferences in each country. The results’ robustness is done by comparing different neural network feedback architectures to improve accuracy through different training.