摘要：By using artificial neural networks [ANN] it was proposed to improve the accuracy of classification of companies within the Bolsa Mexicana de Valores [BMV], specifically the commercial sector compared to multiple discriminant analysis techniques [MDA] and logit models. More than fifty neural architectures were developed, and artificial neural network that resulted was the MLP architecture 6: 12: 2 based learning algorithms retrobackpropagation. The results found with the technique [ANN] showed that this technique has a better prognosis assessment and classification obtained by ADM and logit techniques.
