Prediction of default of enterprises with the use of sugeno-type fuzzy inference algorithm

Abstract

This article describes the methodological approach and experiment to assess the insolvency of company on the basis of fuzzy logic conclusion according to Sugeno-type algorithm. Prior to the applied experiment  the results of scientific researches on the chosen topic are analysed. These works confirmed the judgments about the possibility and meaningfulness of use of adaptive fuzzy neural network on the basis of Sugeno-type algorithm for the diagnosis of default and insolvency. Also it has been described the content, mathematical apparatus of fuzzy logic and algorithm which has been discovered by Sugeno. It has become the basis of methodological approach to forecasting of non-payment of credit by the entities. Comparative analysis of the accuracy of models with different sample sizes, as well as conduction of model optimization has showed that training significantly increases the accuracy of the developed model. The results of the study confirmed a high accuracy of created model. They have also grounded the use of the chosen approach on the basis of fuzzy logic as for the modeling and forecasting of default of legal entities that can be used by banking institutions in their work.