Abstract:

This paper attempts to put forward a variables selection using grey systems theory in order to improve the pattern recognition process realized through neural networks. The pattern recognition facility is used in order to determine each company’s bankruptcy stage through its interconnection with the other companies that are conducting their business in the same field, industry or economy. Compared with a simple model of pattern recognition, our model succeeded in getting a higher accuracy rate. A numerical example, conducted on 61 companies listed in Bucharest Stock Exchange, is also presented