Design And Implementation Of User Credit Score Card Based On Big Data Of Telecom Operators

China's credit information industry started late.At present,the personal credit system has problems such as incomplete data sources,insufficient coverage,and low forecasting efficiency.As a traffic provider for broadband and mobile Internet,telecom operators have accumulated a large amount of user data in business operations,covering all aspects of life,and the credibility is very high.These massive,stable and fastgrowing data are operators.The credit reporting service provides unique conditions.On the other hand,as the scale of telecom services continues to expand and business products continue to innovate,personal credit evaluation for telecom users has become an important part of reducing bad debt risks and increasing business revenue.Aiming at these problems and needs,this paper builds a telecom user credit scoring model based on telecom user data feature selection and modeling,and with reference to the credit scoring mechanism in the financial field.main tasks as follows:(1)Preprocessing of multi-dimensional real business data.Based on the real service data of the telecom operators,the data set is preprocessed,and a fusion algorithm for processing the unbalanced data set is proposed.The algorithm takes into account the advantages and disadvantages of oversampling and undersampling,and implements ENN and ADASYN.The fusion of algorithms.The experimental results show that the fusion algorithm has better processing effect than the traditional unbalanced data set processing method.(2)A feature selection method under a new framework is proposed.Exploratory analysis is carried out on the acquired telecommunication user data.According to the analysis results,the features are single-column processing and feature selection,and the feature selection method of parallel optimization is proposed.After experimental verification,the new method can be used in the case of less features.The most useful information is retained without degrading the performance of the subsequent training model.(3)Design and implement a credit evaluation model based on scorecard form.Using the commonly used machine learning algorithm to construct the telecom user credit evaluation model and carry out experimental comparative analysis,comprehensively consider the predictive ability and interpretability selection logistic regression algorithm to realize the telecom user credit evaluation model,and innovatively apply the financial field scoring card model to In the field of telecommunications,the conversion from model to scorecard is realized.(4)Design and implement a scalable credit scoring prototype system.Combine specific business scenarios and requirements,design multiple functional modules to facilitate subsequent expansion and business applications.In summary,this paper selects the logistic regression algorithm to construct the credit evaluation model for telecom users after data preprocessing and feature processing then applies the bank customer rating card model to the telecom field,providing a feasible idea for the operators to carry out risk control and credit control.