A METHOD OF CORPORATE CREDIT RATING CLASSIFICATION BASED ON SUPPORT VECTOR MACHINE AND ITS VALIDATION IN COMPARISON OF SEQUENTIAL LOGIT MODEL

Abstract

In this study we present a method of multiple discriminant analysis based on support vector machine (SVM) and to apply it to corporate credit rating for financial risk management. In anticipation of improving accuracy ratio of classification of companies with estimated credit rating, we introduce a optimization problem to estimate linear discriminant functions mixed of margin maxitimization and 0-1 integer variables to choose the best set of variables. We show validity of the method to some extent through some comparative analyses between the proposed SVM method and a sequential logit model approach that is one of the most popular statistical models.