2.1 Selection of Data Samples and Their Representativeness
The loan data in this article comes from the daily loan records of Fudeng Small Loan (Sichuan Co., Ltd. The main reasons for selecting this company as a representative are as follows:
- •
The Company Background: The company was established in June 2008, focusing on serving individual industrial and commercial households and micro-enterprises with an annual turnover of less than RMB 60 million. As of the end of 2014, it has provided tens of thousands of individual industrial and commercial households and small and micro enterprises with financing services ranging from a few thousand to millions in Sichuan, Hubei, and Chongqing. The title “Top 100 Competitive Small Loan Companies in China” was issued by the China Federation of Small Loan Institutions.
- •
The Loan Object: The company's loan targets are mainly micro and small enterprises or individual industrial and commercial households. The scope of this study is the credit risk of small loans for small and micro enterprise customers, not for farmers; thus, this company represents this type of small loan company.
- •
The Loan Characteristics: The longest term is 60 months, and the shortest is 1 month. Unsecured loans account for approximately 79.5%. The business is mainly concentrated in Sichuan, Chongqing, and Hubei. The maximum loan amount is 1.2 million yuan, and the minimum is 4882 yuan. Therefore, the loan meets the characteristics of “short term, unsecured, decentralized, and small amount.”
- •
The Data Sources: The data in this article come from the company's actual loan business across the country. Through the company's internal staff, to provide data for research, combined with big data technology, datum is excavated from September 1, 2016 to July 1, 2021. Each loan has 21,299 loan records for each loan attribute (Du, Liu, & Lu, 2021). The period of the data samples is long, and the sample size is large. The data selection is for different small and micro enterprises across the country.
Given that the loan objects of each micro-loan company are random and have the same characteristics, that is, small and micro enterprises with small assets, each micro-loan company has similar loan data registration methods. The database is the same for, for example, Nanchong Meixing Microfinance Company and Bangxin Microfinance Company. The same data recording method and database form of Fudeng Small Loan (Sichuan) Co., Ltd. any are selected in this article. Therefore, the data of Fudeng Small Loan (Sichuan) Co., Ltd. selected in this article are universal, representative, and random.
Experimental methods of BP neural network and support vector machine in MATLAB. In the experiment, the induction method is used. That is, the experiment is carried out with the given representative data, and the results obtained are consistent with this type of data's universality.
Based on the above five reasons, this study analyzes the loan data and attributes of Fudeng Small Loan (Sichuan) Co., Ltd., which is representative of this type of small loan company. Substituting this company's data into the model used in this study can represent my country of this type of small loan.