No abstract

No abstract

This article presents a study on development of credit risk evaluation model using Support Vector Machines based classifiers, such as linear SVM, stochastic gradient descent based SVM, LibSVM, Core Vector Machines (CVM), Ball Vector Machines (BVM) and other. Discriminant analysis was applied for evaluation of financial instances and dynamic formation of bankruptcy classes. The possibilities of feature selection application were also researched by applying correlation-based feature subset evaluator and Tabu search. This research showed that different SVM classifiers produced similar results, including Core Vector Machines based classifier. Yet proper selection of classifier and its parameters remains an important problem.

No abstract

Abstract. This article analyzes the problems of business bankruptcy, and the methods for bankruptcy prediction. This study proposed to join two models, one is the multi-discriminate Z-Score created by Altman, and the other is the Selforganizing maps. We proposed to generate self-organizing maps based on the financial data of public companies that are included in the NASDAQ list. These maps were used for bankruptcy prediction as well as creating classification of financial risk for Lithuanian companies. Comparing the weak results of prediction we accelerated by changing of ratios weights of the Altman Z-Score model. In this way, it can fit to conditions of the Lithuanian conjuncture. Based on the original ratio weights in Altman's Z-Score the results predicting Lithuanian bankruptcy were weak. The weights of Altman's Z-Score model were changed to fit the Lithuanian economic circumstance.