电商时代小额贷款在线信用评估模型的新思路.docxVIP

摘要近年来，网贷领域逐渐发展成为互联网金融的主力军。在国内，以人人贷、拍拍贷为代表的P2P平台和阿里微贷、京东白条领军的电商+金融模式的兴起，无不预示着未来中国互联网贷款平台的大有可为。然而，作为新兴力量，网贷平台存在着不可避免的缺陷，除去法律监管、市场风险等外部因素，平台自身对客户的信用评估系统也尚不成熟。对于网贷领域来说，高违约率带来的巨额违约损失往往是致命打击，截止2015年3月底，因此倒闭甚至“跑路”的平台已超过500家。可见，信用评估模型的精准性恰恰是平台生存、发展链条上至关重要的一环。本文从平台信用风险的角度出发，通过对比P2P平台和电商金融信用风险管控方法的异同，并且考虑到两者的经营方式，仅取其彼此可借鉴之处。以P2P平台为主要研究和试验对象，针对其客户的方式和特点，引入Logistic回归法进行回归分析，以建立新的评估模型，合理评估借款人主体的信用风险。所用运算数据均取自人人贷。“人人贷”网站是国内最早一批提供融资服务的互联网金融平台，并以诚信、公平、透明、创新、高效的特征赢得市场口碑，其网站上每一标的借款人信息披露较为充分，数据收集的效率和准确性有所保证。本文新思路在于：使用传统的Logistic回归模型对新兴平台的借款人信用风险进行评估，同时运用信息增益，得到很好的研究结果和结论，并代入数据进行运用和检验。具体如下：收集“人人贷”网站若干组实验数据，并参考P2P平台、电商金融、商业银行个人风险评估体系指标选取，选取出若干备选指标。通过数据处理、分类和量化的方法，并借助信息增益技术筛选出最终建模指标，继而对建模指标进行回归，得到新的信用风险评估模型。同时将新数据代入新得到的信用风险评估模型中运算，计算出违约比率，进行实际运用的演示和操作。本文最后根据基于回归模型的平台借款人信用风险评估的研究结论，对完善平台借款人信用评估体系提出改进意见。关键词：P2P网贷、信用风险评估、Logistic回归、信息增益AbstractIn recent years, online lending field has gradually become the main force in the development of online lending in China. In order to peer-to-peer lending, pat the loan as the representative of the P2P platform and Ali micro credit, Jingdong-ious leading “e-commerce financial” mode rises, all indicating that future China online lending platform promising. However, as a new power, the online lending platform has inevitable defects; besides legal supervision, market risk and other external factors, the platform of their own credit evaluation system to the customer is not mature. For the online lending field, high default rate of the huge amount of default loss is often a fatal blow, as of the end of march in 2015 , so the collapse or even run away the platform has more than 500. Obviously, the accuracy of credit evaluation model is just the platform surviving and chain of the development of the vital part.From the perspective of platform credit risk, the paper compares P2P platform and financial credit risk management method, and takes the P2P platform as the main research and test object, and the Logistic regression as the main research and test object,