Empirical Study on Indicators Selection Model Based on Nonparametric -Nearest Neighbor Identification and R Clustering Analysis

Table 6

Comparative analysis of 3 models.

Model	Judgement accuracy
	Nondefault customer’s judgment accuracy	Low default loss rate customer’s judgment accuracy	High default loss rate customer’s judgment accuracy	All customer’s judgment accuracy
	Combined model which preserved in this paper	100%	100%	100%	100%
neural network model	99.8%	50%	75%	98.95%
Stepwise discriminant analysis	96.75%	100%	83.33%	96.40%