Serviços Personalizados
Journal
Artigo
Indicadores
- Citado por SciELO
- Acessos
Links relacionados
- Citado por Google
- Similares em SciELO
- Similares em Google
Compartilhar
Revista Ingenierías Universidad de Medellín
versão impressa ISSN 1692-3324versão On-line ISSN 2248-4094
Resumo
PEREZ RAMIREZ, Fredy Ocaris e FERNANDEZ CASTANO, Horacio. LAS REDES NEURONALES Y LA EVALUACIÓN DEL RIESGO DE CRÉDITO. Rev. ing. univ. Medellin [online]. 2007, vol.6, n.10, pp.77-91. ISSN 1692-3324.
In spite of the skepticism of the academic world on the advances of artificial intelligence, the neuronal networks have opened up a field of stock-exchange exploration that has still so much to research. Upon expounding the advantages of the usage of artificial neuronal network (ANN) and its capacity to estimate nonlinear models this article shows the application of the neuronal networks on the quantification of the credit risk. Furthermore, the article carries out a theoretical development of the basic foundations of neuronal networks. In order to present the methodologies of measurements of credit risk, based upon the neuronal networks and to apply the to the data base of a commercial portfolio, it became necessary to elaborate an exploratory analysis of each of the variables and to research the correlation amongst them. The objective of the analysis is to pinpoint some relations for predetermined population groups according to their particular characteristics. Therefore, variables of each client, the credit and the behavior against the variable are crossed default (insolvent and non insolvent): a variable that establishes the classification procedure and determines the necessary averages in addition to establishing the probability of insolvency.
Palavras-chave : Neuronal risk; Basle Committee; credit; networks.