Design And Analysis Of Personal Credit Data Default Mining Model

With the rapid development of economy and the increasing consumption capacity of residents,the personal credit business in the financial industry has developed rapidly.However,illegal activities such as credit default and fraud are increasingly rampant,which seriously affects the financial order of our country.Research on personal credit data and default mining model has become the focus of scholars.Personal credit data is a typical type of unbalanced data.The traditional sample balance algorithm ignores the data noise caused by data edge distribution,and the lack of exploration and analysis of the latest data in the field of credit default.The inadaptability of the mainstream machine learning algorithm in this field also brings difficulties to the application of the model.In view of these urgent problems,this paper has carried out the following work:(1)A radial basis spectrum clustering balance algorithm(RBF-SC)is proposed and designed.In recent years,based on the idea of oversampling,scholars have proposed algorithms such as hierarchical clustering oversampling(WOHC,2019),self-organizing map oversampling(SOMO,2017),Kmeans_SMOTE(2018),etc.,but these algorithms ignore the deviation of the authenticity of oversampling data under the edge distribution of data.In this paper,through boundary decision-making and spectral clustering oversampling,the influence of noise in the balance data set is avoided.Compared with several latest balance algorithms,this method has better data balance ability.(2)Design the RBF-SC method in the background of credit data processing.After processing,the credit balance data is more close to the actual scene to ensure the data quality.On the basis of previous scholars' research,from the platform and customer level,the establishment of credit characteristic analysis system is conducive to assisting credit institutions in business adjustment and enhancing the credit institutions' resistance to risk.(3)An enhanced fusion cascade model(FECM)is proposed.In view of the current situation that the traditional credit model does not match the real scene application well,FECM uses fusion multi-Grained module and enhanced cascade module to build the model.In this paper,the validity of the model is proved by comparing with the existing algorithms on several datasets.In this paper,by comparing the performance of several models on the credit data set,the characteristics of FECM model are deeply analyzed,and the applicability of FECM in the credit problem is proved.(4)A prototype system of FECM credit default analysis based on RBF-SC is developed.Based on the above research results,the framework and route of system research and development are established,and the credit preprocessing module,credit EDA analysis module,data RBF-SC balance module and model building module are designed.Through the research and development of the default analysis prototype system,we can basically achieve the prediction and monitoring of credit default,which has a positive role in promoting the application of data mining in personal credit default prediction.