Credit risk Evaluation Model Based on Self-Organizing Competitive Network

The paper first uses the method of principal component analysis to extract characteristic indexes from 7 financial ratios of management status of the applicants: debt ratio, liquidity ratio, quick ratio, net profit margin of principal business, return on equity, deposit turnover, accounts receivable turnover. Then it is utilized extracted the characteristic indexes to establish two-dimension credit-risk evaluation model based on self-organizing competitive network. The model is used to separate the 80 applicants of a commercial bank of our country into two patterns, which are "credit good" applicants and "credit bad" applicants respectively. The research shows that, the classification accuracy rate of the credit-risk evaluation model based on self-organization competitive network is 98.75%. The result is the same with the literature by Sulin Pang, et al. (2005) which established the credit-risk evaluation model based on BP algorithm. The classification accuracy rate of the BP algorithm is also 98.75%. But from the part of the view of the algorithm, the convergence speed of the algorithm of the self-organization competitive network is fast than that of the BP algorithm so that it is better than the BP algorithm.