Research On Internet Financial Fraud Recognition Based On Large Data

With today’s society has entered the Internet age, the Internet has been closely linked with our lives.Through the changes in people’s habits, the pace of work and industrial development, the Internet is so essential in our lives. P2P--Credit application on network--for its convenience, fast approval, low-threshold characteristics in the credit industry, has quickly find its own living space. But at the same time, it also attracted the attention of industry risk. According to statistics, P2 P network platform credit for every 100 cases refused to loan, there are 16 cases of fraud occur in different ways. They use the characteristics of P2 P network application platform to cheat money. Therefore, the identification fraud activities appeared on Internet financial platform, and predict that whether the customer has a property of fraud come much crucial. Based on the above problems, this paper explored and addressed the problem of fraud appearing in the industry of Internet finance.With the social development, credit-approval in contemporary business is using the user’s property information and other bank information for review. Although the approval process has been great progressed, but it is still not up to expectations for its slow approval and its high threshold. In recent years, the Internet and technological means has been gradually developed, we begin to use data mining and statistical modeling methods to predict customer risk probability, namely the customer’s risk rating. It optimizes the approval process which also save the approval of human consumption, and greatly improves the efficiency of the credit industry. When the "big data" era come, we have put forward higher requirements to the risk control, as well as fraud prediction, and the defense of these behavior. This paper aims to use the advantages of Internet data and big data environments, as well as the use of random forests algorithm, to predict the probability of fraud. In this paper, we use the type of credit score to quantify the probability of fraud which possibly occurred in credit application, and then classify their credit rating.