Research On The Construction Of Credit Rating Model Based On Communication Data

Against the background of long-term dividend,China’s mobile communication industry has encountered the competition of Internet communication,which has greatly affected the profitability of the industry.How to further break the situation in China’s mobile communication industry under the current special environment and background,so as to realize the substantial and positive growth of profits,puts forward higher requirements for its user management level.Credit management is the core link of user management.Its management mode and management level play a decisive role in the user management level of enterprises.However,in the communication industry,operators do not have a special credit management department to regulate the credit management of users,the credit risk management is still in the primary stage,the credit evaluation system is immature.Moreover,from the perspective of the communication industry,although operators have a large amount of user data,they do not make full use of it and convert it into effective resources.Therefore,how to combine user data with credit management and build a more accurate and effective user evaluation system has become particularly important.Therefore,by combing the research results of scholars at home and abroad,this paper analyzes the causes on the basis of understanding the current situation and existing problems of user credit management in the communication industry,and tries to build a credit evaluation system based on user communication behavior based on the data advantages and characteristics of user communication behavior in the communication industry.Specifically,from the historical user communication related data of a communication business outlet,including user consumption behavior,call behavior,shutdown behavior,roaming call behavior and account balance,a total of150000 user record data and 25 indicators are selected as the basic data for the construction of user credit evaluation model.After data preprocessing and variable correlation analysis,the principal component analysis method is adopted,After dimensionality reduction,six principal component factors are finally extracted,and the proportion of each principal component factor and the weight coefficient of relevant variable factors are determined,so as to obtain the calculation formula of comprehensive credit score and establish the credit evaluation model.We further applied the constructed user credit model to the installment purchase business of the company in the second half of 2021.When handling the installment purchase business,customers conducted qualification review in the "1 + 1" way of combining the review of credit rating data model and the credit investigation review of the fund side,which achieved good business application results and reduced the generation of bad debts.