Research On Credit Card Anti-fraud Model

Credit card consumption is a popular form of consumption in today's society,but fraudulent consumption has occurred.These forms of fraudulent consumption include fraudulent use of other people's credit card transactions,forgery of credit cards,phishing scams,etc.,which have caused huge problems for card-issuing banks.Economic losses have brought instability to society.Credit card anti-fraud consumption has become a matter of high concern to banks.The traditional anti-fraud method has high cost,long cycle and low efficiency,and it is difficult to meet the requirements of real-time,which brings great inconvenience to users.The paper collects a number of samples of normal transactions and fraudulent transactions to form a training sample set of anti-fraud transaction identifiers.Each sample includes the number of credit cards,housing status,credit amount,range of years of employment,disposable income,trading time,30 variables such as trading location and transaction volume.The sample training set is used to train the four classifications of random forests,support vector machines,logistic regression and K-nearest neighbors for normal transactions and fraudulent transactions.Each recognition classifier has different recognition accuracy for fraudulent transactions and normal transactions.The efficiency of anti-fraud consumption.If a transaction is identified as a fraudulent transaction by one of the four classification identifiers,the transaction is considered to be suspected of fraudulent consumption,and the transaction is immediately terminated,and zero fraud consumption is achieved as much as possible.The empirical results show that this paper is an anti-fraud consumer identifier built by banks.The recognition rate of fraudulent consumption reaches 98.5% accuracy,which reduces the bank management cost,effectively controls the risk of credit card consumption,and promotes the health of bank credit card business.Stable development.