【作者】 周勇；

【导师】 方壮志；

【作者基本信息】 华中科技大学 ， 金融学， 2016， 硕士

【摘要】 07年以来,P2P网络信贷迅速崛起,以每年数倍的规模增长。但是,整体P2P网络信贷行业还处于初级发展阶段,监管和制度建设都不健全,这就导致了许多问题平台的产生,阻碍了P2P网络信贷行业的发展。而与商业银行不同,大部分的P2P网络信贷公司注册资本较少,管理也较为混乱,未建立完善的信用风险评估机制,对借款人的风险评估很不专业,这就导致了严重的借款人潜在信用风险,从而影响P2P网络信贷行业的发展。本文阐述了P2P网络借贷的特点、风险与发展历程和支持向量机的基本原理,参考商业银行评估个人信用风险的指标与我国P2P网络信贷的特点构建了新的指标体系,用来评估借款人信用风险。在这个指标体系的基础上分析了支持向量机模型在对P2P网络借贷借款人信用评级上运用的可行性,并且运用实际网贷平台数据对构建的多分类支持向量机模型进行了实证研究。经实践证明支持向量机能够很好的预测P2P网络信贷借款人信用风险,具有高准度的分类能力,同时,在数据缺失的情况下,支持向量机也能很好的对借款人的信用风险进行评估,对于P2P网络信贷借款人信用评级具有实用价值。更多还原

【Abstract】 In recent years,P2P lending developed very fast. It increased at a rate of several times per year. But the P2P lending industry is still in a primary stage of development and the regulation and regime are imperfect. Because of these, there appear a lot of problem companies. Different from the commercial banks, most P2P lending companies have a less registered capital, a mess management and lack a system to assess credit risk. These bring a serious potential credit risk and affect the development of the P2P lending industry.This paper states the characteristics, risk and development history of P2P lending and the basic theory of SVM. Then this paper creates an indicators system by studying the indicators system of the traditional bank for evaluating the personal credit risk. And it analyzes the feasibility of evaluating the credit risk of borrower in P2P lending on the basis of the indicators system. It also uses the date that comes from the P2P-lending company to do an empirical analysis. The results shows that SVMs can predict the credit risks of the lenders well, it can exactly classify different lenders. And SVM performs well in the situation where dismissed some data.更多还原