个人信用评估模型比较分析

Comparative analysis of personal credit evaluation models

文章以互联网金融机构客户信贷数据为基础,使用7种不同的数据挖掘方法建立个人信用评估模型,从预测准确率、模型外推性、第二类错误率、预期错误分类成本4个方面评价模型的综合信用评估能力。评价结果表明,使用分类树和K近邻分类算法建立的个人信用评估模型的综合信用评估能力最高;同时发现使用线性和非线性方法建立的模型各有特点,线性分类模型能够对违约客户进行有效识别,而非线性分类模型的预测精度较高。

Based on the customer credit data of internet financial institutions,this paper uses seven different data mining methods to establish personal credit evaluation model,and evaluates the comprehensive credit evaluation ability of the model from four aspects:prediction accuracy,model extrapolation,the rate of typeⅡerror,expected cost of error classification.The evaluation results show that the comprehensive credit evaluation ability of the personal credit evaluation model based on classification tree and K-nearest neighbor classification algorithm is the highest.At the same time,it is found that the models based on linear and non-linear methods have their own characteristics.Linear classification model can effectively identify defaulted customers,while nonlinear classification model has higher prediction accuracy.