Font Size: a A A

The Credit Prediction Of Microfinance Loan Based On Data Mining Technology

Posted on:2017-01-08Degree:MasterType:Thesis
Country:ChinaCandidate:P LiFull Text:PDF
GTID:2279330488982424Subject:Applied Statistics
Abstract/Summary:PDF Full Text Request
With the development of computer technology, the digital era has come. The technological revolution represented by Internet and big data is accelerating people’s steps to enter a new era of development phase-data times following the Agricultural era and industrial Age. The rapid development of Internet technology has produced more than ninety percent of data in the world during the last two years. There exists troubles about how to effectively use the data under the background of the large data. Data mining is a new technology to meet people’s demands for adequate understanding and effective use of the data which contains much information.Recently, along with the Internet online financial services and other new business has grown. Small loans become popular rapidly because of its simple procedure, flexible financing and broad loan range. However, it also brings many problems. For lending institutions:lend loan or not? But the premise of dealing with these problems is to have a thorough understanding of the loan customers’ characteristics, grasp customer’ dynamic, and timely take effective loan scheme. Lending institutions has accumulated a large amount of data in the past business. How to make good use of these data becomes the core to deal with all the problems. Here we can use data mining technology to analyze the huge amounts of data, excavate its spontaneous information and explore secrets of the data, so as to timely seize the market trend, look for valuable customers and optimize the business model.Our article expounds the data mining application in micro loan user data set on the basis of the real and effective data provided by CashBUS. Concrete implementation process are as follows. First of all, use the sampleO function to divided the data sets into training set and test set. Then implement model training on the training set and test model fitting effect on the test set. Because the test result is well known, we can evaluate whether the model is good or not according to the proportion of the predicted results and the real results. We mainly use logistic regression, K-Nearest Neighbor, classification tree analysis and lasso regression. As a result, the test error rates of sequence is 13.8%, 10.98%,18.9%,10.5%. We also carry out the variable selection in logistic regression and the test error rate after selection is 10.8%.The rates has improved than the whole model. Contrasting test error rates of these models, the effect of lasso regression is better.The article not only introduces the basic theories for data mining and credit assessment, but also makes a profound analysis on the research status of the credit evaluation. Data information discovery has brought unprecedented opportunities to the enterprises and even the government. However, how to make effective use of the data become a difficult problem. This article takes the character analysis of user data in the microfinance loan area for example, showing readers how to analyze the data to discover information of the real enterprise data. The article also presents the principles of data mining technology in detail in order to make it easier for future reference. We can use the rich customer data to constantly modify the model parameters, change the model, so that we can reduce the loan risk to a certain extent, and improve the level of loan business. The ideas of dealing with problems in the article, also provide ideas for the future social development:we can use the "big data" and "algorithm" to make business intelligent, so as to provide good ideas for business decisions and enterprise development.
Keywords/Search Tags:Data mining, Credit evaluation, Microfinance loan data, Logistic regression, K-Nearest Neighbor, classification tree, Lasso regression
PDF Full Text Request
Related items