PREDICTION OF CHANGES OF BANKRUPTCY CLASSES WITH NEURO-DISCRIMINATE MODEL BASED ON THE SELF-ORGANIZING MAPS

Abstract

It is suggested in this paper generating SOM as one that could be applied for forecasting of bankruptcy classes for other than trained companies. The various aspects of the proposed Neuro-discrimi-nate model based on the multiple discriminate analysis, supervised learning neural network and self-orga-nizing maps are analyzed. There is compared how accuracy of prediction changes executing algorithm with different discriminate models of bankruptcy: Springate, Zmijewski and Shumway.