P2P网贷中借款人的信用风险评估研究

发布时间：2018-03-25 20:10

本文选题：P2P网贷　切入点：信用评价　出处：《首都经济贸易大学》2017年硕士论文

【摘要】：P2P网贷全称是P2P网络借贷。P2P是英文单词peer to peer的缩写,是“点对点”或“个人对个人”的意思。P2P网贷是依托互联网背景发展而来的一种新兴的借贷模式。这种借贷交易没有最低交易金额的限制,依托网络就可以迅速完成,大大的节省了交易的时间成本。截止到2016年底,国内已有四千多家P2P网络信贷平台,其发展速度与规模都让人惊叹不已。这种准入门槛低、无抵押无担保、简单便捷的互联网金融服务模式,给了传统金融行业很大的冲击,但同时也满足了特定人群的需求,解决了很多个人和小微企业贷款难和融资难的困境,还能让更多的工薪阶层参与到金融理财的队伍中来,合理的利用了社会资源,蕴含着巨大的市场潜力。但在满足广大投资者与借款者需求的同时,这种网络信贷业务,也随着发展进程慢慢的呈现出了很多隐藏的问题。截至2016年年底,国内出现问题的P2P网贷平台已高达数百家,平台倒闭与平台责任人跑路的现象屡屡发生。由于P2P行业在国内起步较晚,对借款者信用的评价与管理技术不成熟,尚未有对借款人信用进行评价的有效方法。鉴于此,本文重点对平台借款人信用评价的方法做了深入研究。文章首先对国内外典型的P2P平台运营模式与特点进行了深入剖析与探讨,借鉴国外较成熟的信用评价方法,并结合国内P2P网贷平台的特点,建立了相应的P2P借款人信用风险评价指标体系。依托该指标体系建立相应的因子分析-支持向量回归(FA-SVR)模型,然后借助libsvm软件包对模型进行训练与优化。该模型可以根据网贷平台提供的借款人信息对其信用风险进行评估,平台经营管理者可以以此为参考,对平台借款人进行有效的管理,对于投资者,该模型也可以帮助其较好的识别与选择借款人资信情况较好的投资项目,在降低投资风险的同时获得较好的收益。本文利用python爬虫技术抓取国内具有代表性的P2P网贷平台——人人贷上的借款人相关信息。首先借助因子分析方法对借款人进行粗略打分,之后用这些数据对支持向量回归模型进行训练与优化,并用部分借款人信息对模型预测的准确率与实用性进行了验证,同时借助gridregression函数对模型进行优化与改进。结果表明经过优化的FA-SVR(因子分析-支持向量回归)模型相对于未经优化的因子分析-支持向量回归模型具有更好的收敛性和更高的预测能力,能更好地用来对借款人的信用进行评价,从而能够对P2P网贷平台的运营风险很好的把控,对平台投资者在投资项目的选择上有了更好的参考。
[Abstract]:P2P network loan full name is P2P network loan. P2P is the abbreviation of the English word peer to peer. P2P loan is a new lending model developed from the background of the Internet. There is no limit on the minimum transaction amount, and the loan transaction can be completed quickly by relying on the network. By the end of 2016, there were more than 4,000 P2P network credit platforms in China, and their development speed and scale were amazing. This kind of low entry threshold, unsecured and unsecured, The simple and convenient Internet financial service model has given a great impact to the traditional financial industry, but at the same time it has also met the needs of specific groups of people, and solved the plight of many individuals and small and micro enterprises with difficult loans and financing. It can also allow more working class to participate in the financial management team, make rational use of social resources, and contain huge market potential. But while meeting the needs of investors and borrowers, this online credit business, With the development process, many hidden problems have emerged. By the end of 2016, there had been hundreds of P2P loan platforms in China. Due to the late start of P2P industry in China and the immature technology of evaluating and managing borrowers' credit, there is no effective method to evaluate borrower's credit. This paper focuses on the credit evaluation methods of platform borrowers. Firstly, this paper analyzes and discusses the typical P2P platform operation mode and characteristics at home and abroad, and draws lessons from the more mature foreign credit evaluation methods. Combining with the characteristics of domestic P2P network loan platform, the corresponding credit risk evaluation index system of P2P borrowers is established. Based on the index system, the corresponding factor analysis-support vector regression (FA-SVR) model is established. Then the model is trained and optimized with the help of libsvm software package. The model can evaluate the credit risk according to the information of the borrowers provided by the network loan platform, and the platform managers can use this model as a reference. For the investors, the model can also help them to identify and select the investment projects with better credit standing. In this paper, we use python crawler technology to capture the relevant information of borrowers on peer-to-peer lending, a representative P2P network loan platform in China. Firstly, we use factor analysis method to score the borrowers roughly. Then the support vector regression model is trained and optimized with these data, and the accuracy and practicability of the model prediction are verified with partial borrower information. The results show that the optimized FA-SVR( Factor Analysis-support Vector regression) model has better convergence than the unoptimized FA-SVRmodel. Sex and higher predictive power, It can be better used to evaluate the credit of the borrower, so it can control the operation risk of P2P network loan platform, and provide a better reference for the platform investors in the choice of investment projects.
【学位授予单位】：首都经济贸易大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：F832.4;F724.6

【相似文献】