Forecasting of changes of companies financial standings on the basis of self-organizing maps

The multivariate discriminate models have been used in area of bankruptcy analysis for many years. In this paper we suggest to conjunct the principles of traditional discriminate bankruptcy models with modem methods of machine learning. We propose the forecasting model based on Self-organizing maps, where inputs are indicators of multivariate discriminate model. Accuracy of forecasting is improved via changing weights with supervised learning type ANN. We've presented results of testing of this model in various aspects.