基于数据挖掘技术的微额借款用户的信用预测

发布时间：2018-06-20 00:07

本文选题：数据挖掘 + 信用评估　；参考：《华中师范大学》2016年硕士论文

【摘要】：随着计算机技术的发展,数字化时代已经到来。以互联网、大数据为代表的这场技术革命正在引领人们加速进入继农业时代、工业时代之后的又一个新的发展阶段—数据时代。世界上百分之九十以上的数据都是近两年产生的,然而庞大数据的背后是如何才能有效利用它们的问题。数据挖掘就是为满足人们对于数据中蕴含信息的充分理解和有效运用而发展起来的一门新兴技术。近几年,伴随着互联网发展起来的还有线上金融服务等新型业务,其中小额贷款以其程序简单、资金灵活、贷款范围广等优势迅速走红,相应的也给贷款机构带来了诸如：贷不贷,贷多少的问题。要处理好这些问题就要尽可能全面的了解贷款客户的各项特征,把握客户动态,及时拟定有效的业务方案。贷款机构在过去的业务实施过程中积累了大量数据,如何利用好这些数据成为解决问题的关键。这里就可以运用数据分析技术,对数据进行分析,及时把握市场动向,优化业务模式。本文就是基于微额速达金融服务公司提供的真实有效数据,详细阐述了数据挖掘在微额借款用户数据集中的应用。具体实施过程是：在R软件中,首先利用sampleO将数据集分为训练集和测试集,然后在训练集上进行模型训练,在测试集上测试模型拟合效果。因为测试集数据结果是已知的,所以可以根据预测值与真实值的比例评价模型好坏。我们这里主要运用了逻辑斯谛回归、K最邻近、决策树和lasso回归进行建模分析,用它们得到的测试错误率依次为：13.8%、10.98%、18.9%、10.5%。期间我们还对logistic回归进行了变量选择,选择后的测试错误率为10.8%,比全模型有所提升。对比这几个模型的测试错误率,lasso回归的效果要稍微好一点。文章不仅对数据挖掘和信用评估的基本理论知识作了介绍,也对信用评估的研究现状做了深刻的分析说明。大数据时代的来临,给企业甚至是政府都带来了很好的发展的契机,然而对数据的有效利用却成为难题。本篇文章就以微额借款用户数据的人品分析为例,向读者展示了如何运用数据挖掘技术对真实的企业数据进行信息发现的过程。文章还对用到的挖掘技术的原理进行了详细介绍,方便读者以后的研究工作。通过运用丰富的客户数据,不断地修改模型参数,优化算法模型,能够在一定程度上降低贷款风险,提高贷款服务质量。文章中的问题处理办法,也给今后的社会发展提供了借鉴：我们可以通过“大数据”和“算法”让商业变得智能,从而为商业决策、企业发展提供好的思路。
[Abstract]:With the development of computer technology, the digital age has come. The technological revolution, represented by the Internet and big data, is leading people to enter a new stage of development after the agricultural age and the industrial age-the data age. More than 90% of the world's data have been generated in the last two years, but behind the huge numbers is the question of how to use them effectively. Data mining is a new technology developed to satisfy the full understanding and effective application of information contained in data. In recent years, along with the development of the Internet, there have also been new types of business such as online financial services. Among them, micro-loans have quickly become popular because of their advantages of simple procedures, flexible capital, wide scope of loans, and so on. It also brings problems such as not lending and how much to lend to lenders. In order to deal with these problems, we must fully understand the characteristics of loan customers, grasp customer dynamics, and formulate effective business plans in time. Loan institutions have accumulated a large amount of data in the past business implementation process, how to make good use of these data become the key to solve the problem. Here can use data analysis technology, data analysis, timely grasp market trends, optimize business models. This paper is based on the real and effective data provided by Microfinance Services Company, and describes the application of data mining in the data set of micro-loan users in detail. In R software, the data set is divided into training set and test set by sampleO, then model training is carried out on the training set, and the model fitting effect is tested on the test set. Because the results of the test set are known, the model can be evaluated according to the ratio of the predicted value to the real value. We mainly use the logicality regression to model and analyze the nearest neighbor, decision tree and lasso regression. The test error rates obtained by them are: 1: 13.8and 10.98 / 18.9 / 10.5 respectively. We also selected variables for logistic regression. The error rate of the selected test was 10.8, which was higher than that of the whole model. The test error rate of these models is slightly better than the lasso regression. This paper not only introduces the basic theoretical knowledge of data mining and credit evaluation, but also makes a profound analysis of the present situation of credit evaluation. With the advent of big data era, enterprises and even the government have a good opportunity to develop, but the effective use of data has become a difficult problem. This paper takes the personal character analysis of micro-loan user data as an example to show readers how to use data mining technology to discover the information of real enterprise data. The paper also introduces the principle of mining technology in detail, which is convenient for readers to study in the future. By using abundant customer data, constantly modifying the model parameters and optimizing the algorithm model, the loan risk can be reduced to a certain extent and the loan service quality can be improved. The problem solving methods in this paper can also be used for reference for the future social development: we can make the business intelligent through "big data" and "algorithm", so as to provide a good way of thinking for business decision making and enterprise development.
【学位授予单位】：华中师范大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：F832.4

【相似文献】