独创性声明
本人声明所呈交的论文是我个人在导师指导下进行的研究工作及取得的研
究成果。尽我所知，除了文中特别加以标注和致谢的地方外，论文中不包含其他
人已经发表或撰写过的研究成果，也不包含为获得北京工业大学或其它教育机构
的学位或证书而使用过的材料。与我一同工作的同志对本研究所做的任何贡献均
已在论文中作了明确的说明并表示了谢意。
签
日
名：
期： 2021年 5月 30日
关于论文使用授权的说明
本人完全了解北京工业大学有关保留、使用学位论文的规定，即：学校有权
保留送交论文的复印件，允许论文被查阅和借阅；学校可以公布论文的全部或部
分内容，可以采用影印、缩印或其他复制手段保存论文。
（保密的论文在解密后应遵守此规定）
签
名：
日
日
期： 2021年 5月 30日
期： 2021年 5月 30日
导师签名：

摘要
摘要
随着大数据时代的到来以及互联网金融的快速发展，新兴的金融科技公
司依托于其在数据上的积累和科技上的优势，在小额信贷方面以高效、快速
的经营模式迎合了个人和小微企业贷款的需求，快速抢占着商业银行的市场，
对商业银行的业务产生了很强的冲击，同时商业银行自身还面临着不良贷款
率高企的问题。为了提升自身竞争力，降低不良贷款率，商业银行需要解决
传统信贷风控中信息不对称、时效性差、决策效率低和成本高等问题，商业
银行通过大数据风控的应用，实现用户信贷风险状况的精准画像，提升信贷
业务的决策效率和准确度，打造商业银行的核心竞争力。
本文以互联网大数据环境下ZY银行信贷风控为研究方向，分析该银行现
有风控模式存在的问题以及从大数据风控角度出发如何摆脱当前面临的困
境。通过多维度数据的收集和清洗提升数据质量，为大数据决策提供基础保
证，根据不同变量在模型中的显著性选取对风控决策有重要意义的变量进入
模型，实现对用户信贷准入和授信额度的判断。通过信用评分卡的输出实现
多因素决策的量化展示，利用不同阶段评分卡的创建和使用对用户贷前准入、
贷中风险监控和贷后催收阶段的风险进行识别和判断。市场不断变化，风控
永不停止，风控模型需要不断的迭代更新，本文通过模型的监控指标分析判
断模型是否需要调整。最后通过ZY银行大数据风控平台的建设分析了平台化、
服务化方式对于模型的应用和信贷业务快速决策带来的价值。
通过大数据风控平台的建设和大数据风控的应用，ZY银行在不良贷款率
上有了明显的下降。基于互联网的信贷业务实现了高效的发展，提升了信贷
决策效率，降低了信贷决策成本。
关键词：大数据；商业银行；信贷风控；互联网
-I-

Abstract
Abstract
With the rapid development of big data and Internet-based finance, the emerging
financial technology companies have quickly seized the market of commercial banks
by relying on their data accumulation and technological advantages to cater to
individuals and small and micro businesses with efficient and fast business models,
which has had a huge impact on the business of commercial banks. At the same time,
commercial banks are also facing the problem of high non-performing loan rates. In
order to improve their competitiveness and reduce the rate of non-performing loans,
commercial banks need to solve the problems of information asymmetry, poor
timeliness, low decision-making efficiency and high cost in traditional credit risk
control. Through the use of big data risk control, commercial banks can achieve
accurate user portraits, improve the efficiency and accuracy of credit business decision-
making and enhance the core competitiveness of risk control.
This thesis takes ZY bank’s credit operations risk control under the big data
environment as the research direction. It analyzes the bank’s existing risk control model
problems and how to solve and optimize the current dilemma from the perspective of
big data risk control. In the application of big data, ZY bank improves data quality by
collecting multi-dimensional data and doing data cleaning, which is the basis of big
data decision-making. According to the significance of different variables in the model,
variables that are important for risk control decision-making are selected to enter the
model, so as to achieve user credit access and credit judgment of quota. The output of
the credit score card is used to realize the quantitative output of multi-factor decision-
making, and the creation and use of scorecards at different stages are used to identify
and judge the risks of users in the Loan origination, Loan maintenance, and delinquency
management stages. The market is constantly changing, and risk control never stops.
The risk control model needs to be updated iteratively. This thesis analyzes the
monitoring indicators of the model to determine whether the model needs adjustment.
Finally, through the construction of ZY bank's big data risk control platform, the value
of platform-based and service-based methods for model application and rapid credit
business decision-making is analyzed.
Through the construction of the big data risk control platform and the application
of big data risk control, ZY Bank has significantly reduced its non-performing loan rate
in recent years. With the support of big data, The Internet-based credit business has
-III-
。。。以下略