Evaluation Of The Effect Of The Personal Loan Default Model After Adding The Characteristics Of The Guarantee Network

In recent years,the scale of banks' personal loans has continued to grow,and the risks they face are also increasing.How banks use data to predict personal loan defaults more scientifically has become the key to current bank risk prevention in this area.With the continuous improvement of our country's credit investigation system,the continuous enrichment of Internet scenarios,and the development of new technologies such as big data and cloud computing,the data dimensions of customer credit evaluation continue to expand,and the selection of characteristic variables and model methods continue to evolve and optimize.There are two trends in general.One is that the characteristic dimension of credit evaluation has broadened from focusing on the person's own characteristics to the fusion of the person's own characteristics and the characteristics of social relations;the second is that the evaluation model is more enriched.Judging from the existing literature,there are relatively abundant researches on corporate clients along these two trends,while research on individual clients is rarely seen,especially when selecting "guarantee relationship" as an indicator of "social relations" in personal evaluation.Based on the previous literature on personal loan default prediction,this paper analyzes the nature of personal loans in combination with the actual situation of commercial banks,and puts forward the idea of using the information of guarantee network and implicit guarantee network to predict loan default.First,build a social network based on the personal loan guarantee information of commercial banks,and then go a step further,use the relevant evolutionary algorithm to predict the implicit guarantee network based on the structural similarity between individual loan customers and related indicators of individual similarity.Then we combines the actual network connection and the implicit network connection and mines the factors that can be used to predict the network that affects the default status of personal loans,thereby improving the effect of predicting the default of personal loans.We believe that the closely connected individuals in the guarantee network have mutual influence.This article decomposes the mutual connection and influence of the nodes in the guarantee network into three parts and combines the concept of centrality to measure the surrounding people's influence through a certain algorithm.Two features are created and added to the forecasting model as an indicator.Finally,according to the actual data of the personal loan information records of commercial banks and related personal loan guarantee information,this paper conducts an empirical analysis of the personal loans of commercial banks after the division,and at the same time adds the influence indicators extracted from social networks into the machine learning prediction model for preliminary verification The above conclusions are reached.