CREDIT RATING OF COMPANIES WITH NEURAL NE TS

Because of the increasing number of company insolvencies and the corre sponding loan losses, there is a need for new methods of credit rating with the ability to predict bankruptcies early. Employing Artificial Neural Nets (ANN) an indicator could be developed which identifies a c ompany crisis early, objectively and with a high reliability. The anal ysis is based on a large and representative data set. Out of a great n umber of predictor variables that combination of variables has been de termined best classifying companies into failed and non-failed classes . For this purpose a backpropagation net and pruning algorithms have b een used. The results of the developed ANN are plausible and easy to i nterpret. In credit investigation ANN can be adapted to the parameters of a bank for optimal results. As a conclusion ANN are an effective t ool to predict commercial failures.