Abstract
With the rapid development of big data technology, the personal credit evaluation industry has entered a new stage. Among them, the evaluation of personal credit based on mobile telecommunications data is one of the hotspots of current research. However, due to the complexity and diversity of personal credit evaluation variables, in order to reduce the complexity of the model and improve the prediction accuracy of the model, we need to reduce the dimension of the input variables. According to the data provided by a mobile telecommunications operator, this paper divides the data into a training sets and verification sets. We perform correlation analysis on each indicator of the data in the training set, and calculate the corresponding IV value based on the WOE value of the selected index, then binning data with SPSS Modeler. The selected variables were modeled using a logistic regression algorithm. In order to make the regression results more practical, we extract the scoring rules according to the results of logistic regression, convert them into the form of score cards, and finally verify the validity of the model.
Highlights
Credit investigation refers to the collection, sorting, preservation, and processing of credit information of natural persons, legal persons and other organizations in accordance with the law, and the provision of services such as credit reports, credit evaluations, and credit information consultations [1]
Due to the complexity and diversity of personal credit evaluation variables, in order to reduce the complexity of the model and improve the prediction accuracy of the model, we need to reduce the dimension of the input variables
We perform correlation analysis on each indicator of the data in the training set, and calculate the corresponding IV value based on the WOE value of the selected index, binning data with SPSS Modeler
Summary
Credit investigation refers to the collection, sorting, preservation, and processing of credit information of natural persons, legal persons and other organizations in accordance with the law, and the provision of services such as credit reports, credit evaluations, and credit information consultations [1]. A credit score card model based on historical data and using statistical methods to assess customer risk begins to emerge [6]. Given that the complexity and diversity of credit evaluation variables, and the accuracy of the logistic regression model, first, the weight of evidence-information value (WOE-IV) method will be employed to select the variables [17]. The dimensionality reduction variables and the logistic regression method will be utilized to record data on the customer behavior of a communication operator in China, aiming to establish a statistical analysis model for personal credit evaluation to differentiate between “good” customers and “bad” customers. For customers with poor credit records, increasing control can effectively reduce the risk of arrears and bad debts; for valued customers with good credit, some preferential packages and other services should be launched to attract more users to come back again, thereby enhancing the competitiveness of the enterprise [18]
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
