Empirical Study on Indicators Selection Model Based on significant Discrimination and R Clustering Analysis

Small enterprises play the important role in pushing China's economic progress, but keeping on facing the difficulty in financing the loans. Establishing a reasonable credit evaluation indicators system is one of the keys to implement accurate credit evaluate to small enterprises. Regardless of the evaluation method being used, with unsuitable indicators system, it is impossible to obtain reasonable credit evaluation results. By the application of logistic regression significant discrimination and R clustering analysis, a small enterprises credit evaluation indicators system is established. The credit evaluation system established in this paper is capable of significantly discriminating default samples from non-default ones and can effectively avoids duplicate information. The result of empirical study shows that the credit evaluation indicators system established in this paper is able to reflect 83.47% of original information with 22.22% of original indicators.