互联网金融平台中高违约风险用户识别算法

在我国金融科技不断创新的背景下，互联网金融平台中通过网络分析技术开展用户风险识别已经成为当前的热点技术发展方向。以某互联网金融平台的用户交易数据为对象，通过分析其中借贷逾期违约的传播行为，提出通过传播特征构建模型算法识别互联网金融平台的高风险用户。在构建基于阈值传播和随机传播的SIS模型和SIR模型基础上，将模型转换为可评价用户风险值的算法，并进一步与实际违约数据进行验证对比。对比结果显示在前5%和10%高风险群体划分条件下，算法具有较高的召回率和良好的结构关联性。

Identification algorithm of high breaching risk member for Internet financial platform

At the background of increasingly developed financial technology, identifying of high-risk members through network analysis method on internet financial platform has become a newly hot research area. In this paper, we studied loan transaction data of an internet financial platform, and identified the high-risk members by analyzing the propagation behavior of the loan network. We established the SIS model and SIR model based on threshold propagation and random propagation, respectively. After that, we generated an algorithm to evaluate the users'' risk value. Furthermore, we compared it with the actual defaulting data. In terms of the top 5% and 10% high risk group division, the results show that we can achieve high recall rate and good structural correlation with the algorithm.