Research On Anti-fraud Of Network Lending Based On Machine Learning Algorithm

With the continuous progress of information technology,the rapid development of economy,the gradual change of people's consumption concept and form,the relationship between the Internet industry and the banking and financial industry is increasingly close.All kinds of Internet financial companies are also emerging.The emergence of Internet finance not only breaks the traditional way of bank lending,enables people to directly realize online financing,but also makes financial transactions more convenient and efficient.And it also spawns many Internet small loan companies and Internet Financial platforms.Various Internet lending businesses continue to rise.However,as the whole industry is still in the rising stage,there are a mixture of domestic Internet lending platforms.Financial fraud means emerge in endlessly,and the technical means of risk control departments are imperfect.So how to do a good job in risk control of the Internet financial industry is also a big problem.On the basis of the former antifraud of network loan,this paper combines the theoretical analysis and empirical research.Using the collected user credit information to extract the behavior characteristics of normal users and fraudulent users,the paper compares and analyzes the normal users and fraud users at different levels.For the difference of users,we use machine learning algorithm such as support vector machine,random forest and GBDT model to carry out data mining comparative study.We explore the effect of the model from the accuracy rate,recall rate,accuracy rate and AUC value of model recognition,and analyze that different models have advantages in different dimensions.For the data in this paper,the overall performance of random forest and GBDT should be more excellent.Relevant personnel can select the appropriate model according to the actual needs.