Clasificación de hogares pobres en la región andina de Venezuela usando redes neuronales

This research is aimed to build a classifier for poor houses in the Andean region of Venezuela using artificial neural networks. For this, it was necessary to develop a multidimensional poverty indicator based on the Alkire and Foster methodology, which was modified and developed based on the shortcomings and not on the achievements. This indicator considers that poverty depends on several aspects, and six dimensions are proposed to measure it: housing conditions, housing construction materials, adaptation of services, home economics, education and health. When examining the classification obtained with backpropagation algorithm-based neural network models, it was found that the classification errors are very small, almost perfect, while the models based on radial-based functions presented higher classification errors.