Título: A Neural Network Based Solution for the Credit Risk Assessment Problem
Autores: Vasconcelos, Germano C.; Adeodato, Paulo J. L.; Monteiro, Domingos S. M. P.
Resumo: The automation of decision making in financial markets is one of the major application areas of neural networks. Risk analysis is one of the problems where the technique has been efficiently applied. This paper investigates a solution to a credit analysis problem in a rather peculiar environment, characterized by a stabilized economy but subject to a high interest rate, namely the Brazilian market. A neural network based credit scoring system has been developed for the retail business in Brazil and its performance has been evaluated against that attained by a traditional discriminant analysis system. Extensive experimental results carried out with a database of 18,000 consumers of a leading Brazilian supermarket chain clearly indicate that a better solution is found with the connectionist based system.
Artigo em pdf: 4cbrn_063.pdf
Arquivo BibTex: 4cbrn_063.bib