Research On The Value Mangagement Of Data On Credit Risk Control Layoutunder The Backgroud Of Privacy Calculations

As a new production factor,data is becoming the key factor to realize the digital economy.The method of data value management such as circulation sharing,value mining plays an important role on the transformation of digital.As one of the key productions on credit risk control which is the key finanical business,data plays a decisive role on risk assessment and risk control.With the issue of the protection law of personal information,the safty of data security,the regulations on the administration of personal credit,the security of data and personal information are being pay more and more attention.Firstly,this paper shows the theories related to data value,credit risk control and privacy computing,which provides a theoretical basis for the research.Data value management is propsed which includes data circulation,data fusion and data value mining.Then making the deeply analysis under the backgroud of privacy caclulations.Secondly,it elaborte the detailed data source of credit risk control involved the business processes before,during and after loan.Also analyzing the problems of data value management in credit risk control business combined with the key points of personal information;Finally,based on the method of case analysis and experimental comparison,taking the two actual scenarios of anti fraud modeling,blacklist and so on as the case,it reaserches that how to make the maagement of data value in the process of credit risk control under the backgroud of privacy caclulations.Based on the technology of vertical federal learning,it compared the modeling effects between the traditional model and the model under vertical federal learing.Analysising the experiment data,we can get the conclusion that the two model effects have no big difference which adopts xgboost algorithm and takes AUC and KS as model evaluation indicators.At the same time the model can meet the needs of business.While the model based on vertical federal learning can avoid the leakage of data and garuntee the circulation of data value.By combining diversified data sources,we can improve the effect of credit risk control model and improve the avlue of data.