大数据背景下消费信贷风险因素分析

发布时间：2019-02-14 20:02

【摘要】：相对于传统的信用卡业务,消费信贷业务一般具有单笔授信额度较小,审批时间短,无需抵押或质押,贷款期限较短等特点,因而更加灵活,受到广大商家尤其是电商平台以及广大消费者的欢迎。本文介绍了消费信贷的发展背景以及发展消费信贷对于我国经济发展的重要意义,然后通过进行国内外文献综述,梳理出消费信贷风险控制的发展方向。通过查阅资料我们发现国外的征信模式主要有同业征信、公共征信和市场征信三种模式,利用大数据进行消费信贷风险控制主要是基于市场征信。而国内利用大数据进行消费信贷风险控制的公司主要为BAT(百度,阿里和腾讯),其中以阿里尤为典型,也取得的不错的成效。对于没有数据量积累的消费信贷公司,我们也可以通过多种途径来收集消费信贷数据,并结合两大类个人信用评分模型-统计模型和非统计模型,来对消费者的消费行为进行刻画,为风险控制提供依据。本文利用“一起好”“365易贷”和“易融恒信”的数据,利用神经网络方法处理得到的“好客户”和“坏客户”数据,来分析利用大数据判别申请者信用状况的可行性。实证表明,利用大数据手段并结合神经网络模型判别客户的“好”与“坏”,准确率可以达到80%左右;同时,文章分析了信贷奉献中各个因素的重要性。从而达到了本文的预期效果。总体来讲,本文的创新之处在于,利用网络开源数据,结合大数据手段,从更多地维度去刻画消费者的信贷行为,探索出消费信贷的特征,为消费信贷业务开展的风险控制提供了全新的思路,具有良好的应用前景。
[Abstract]:Compared with the traditional credit card business, the consumer credit business generally has the characteristics of small credit line, short approval time, no mortgage or pledge, short loan period, and so on, so it is more flexible. By the majority of businesses, especially the e-commerce platform and the majority of consumers welcome. This paper introduces the background of the development of consumer credit and the significance of the development of consumer credit to the economic development of our country, and then, through the literature review at home and abroad, sorts out the development direction of the risk control of consumer credit. We find that there are three modes of credit reporting in foreign countries: peer credit, public credit and market credit. Big data is used to control consumer credit risk mainly on the basis of market credit. The main domestic companies using big data to control consumer credit risk are BAT (Baidu, Ali and Tencent), which is especially typical of Ali and has achieved good results. For consumer credit companies with no accumulation of data, we can also collect consumer credit data through a variety of channels, and combine two main types of personal credit scoring models-statistical model and non-statistical model. To describe the consumer's consumption behavior, to provide the basis for risk control. This paper analyzes the feasibility of using big data to judge the credit status of applicants by using the data of "good together" 365 easy loan "and" Yi Rong Heng Xin ", and using the neural network method to process the data of" good customers "and" bad customers ". The empirical results show that the accuracy rate of judging customer's "good" and "bad" by using big data method and neural network model is about 80%. At the same time, the paper analyzes the importance of each factor in credit offering. Thus, the expected effect of this paper is achieved. In general, the innovation of this paper lies in the use of online open source data, combined with big data means, to depict consumer credit behavior from more dimensions and explore the characteristics of consumer credit. It provides a new idea for the risk control of consumer credit business, and has a good application prospect.
【学位授予单位】：天津大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：F832.4

【相似文献】