Application Research On Artificial Intelligence Algorithm In Credit Rating Of Telecom Operator Users

As market economy enters a new stage of high quality development in China,the construction of the credit system has become an important trend that is inevitable.Whatever it is an institution,an enterprise or an individual,credit plays an increasingly important role in production and management and daily life.Nowadays,the Internet economy still keep developing,and credit derivatives emerge in an endless stream.P2P and online shopping platforms have released complex consumer credit products.However,as emerging platforms are still in the initial stage of expanding user scale,credit risk has become a key issue for the company's development.The rise of the"Internet+" model has not only made the competition among the three major telecom operators hot,but also made some Internet companies join the competition.In order to obtain more user resources,telecom operators have to get lower the threshold for access,but this behavior causes users to owe money to the network frequently,which is also an important reason for the company to cause losses.It can be seen that the research and resolution of credit risk issues is imminent.In recent years,artificial intelligence technology has been widely used in many fields.It also provides a new solution for big data problem analysis.With the advantages of natural user scale,telecom operators have the characteristics of wide source of scenes,strong real-time performance,real reliability and wide coverage,which provide data support for artificial intelligence in solving problems in credit risk rating.Therefore,artificial intelligence also brings new ideas to solve the problem of telecom operators' credit rating.This paper first briefly introduces the research background,research significance,research status in China and abroad,research purposes,innovation points,research methods and paper structure of telecom operator user credit rating issues.Secondly,the basic concepts of artificial intelligence algorithm,the basic principles of the model,and its advantages and disadvantages are discussed in detail.Then,using the dataset of the telecom operator user in a city in Jiangsu Province as an example,the Python language is used for data exploration,cleaning,feature engineering,etc.The cleaned data set is applicable for Logistic Regression,Support Vector Machine,Multi-Layer Perceptron,Decision Tree,Random Forest,XGBoost and stacking.The results of the above model fitting were evaluated and compared by selecting the accuracy,macro precision,macro recall,macro F1,ROC curve and AUC.Finally,based on the results of empirical research,the credit rating of telecom operators is summarized and forecasted.This paper models telecom carrier data that includes multiple dimensions such as user identity,behavioral characteristics,spending power,social relationships,and credit history.The empirical results of the comparison model are:current weapon in data modeling competition-stacking based on XGBoost has better effect in solving the credit rating problem of telecom operators than using a single model.From the performance of the test set,the use of this model fusion can not only distinguish a certain percentage of credit rating users,but also avoid excessive "mistaken" users of normal credit ratings.At the same time,the model fusion includes cross-validation,which can effectively prevent the model from over-fitting,and the result is more general.Therefore,through research,this paper believes that the model fusion can effectively assist telecom operators to predict and identify users with different credit ratings in a timely manner,reduce the company's losses,and provide personalized services for users of different credit ratings to improve user satisfaction.