Application Of Combination Classifier In Credit Card Fraud Detection

Credit card fraud refers to the deliberate use of forged other people's credit cards,theft of other people's credit cards for personal consumption,or the use of credit cards to cash in violation of the law.Credit card business is one of the main business of banks.In recent years,it has become an important profit growth point.It is an intermediary business with great potential which is unanimously recognized by foreign financial industry.In 1985,China's first credit card was issued by the Bank of China.In the 21 st century,the credit card business grew rapidly and the total issuance continued to rise.At present,the ratio of debit card issuance to credit card issuance is about 9:1.With the widespread use of credit cards,the number of illegal acts of credit card theft is also on the rise.Credit card theft has caused great economic loss,time loss and social credit loss to issuing banks and consumers who normally hold credit cards.Credit card issuers in China are paying more and more attention to the problem of credit card fraud and monitoring the management of credit card transaction records,trying to reduce the economic losses of consumers and businesses,prevent credit card theft in time,and effectively control the incidence of credit card fraud and crime.With the popularity of credit cards,various types of credit card fraud also follow.In order to help credit card users reduce losses and help credit card issuing agencies identify credit card fraud,this paper attempts to use logistic regression classification method in data mining combined with adaptive enhancement algorithm,through the collected data sets of credit card transactions,trying to establish a new perspective.A classification model can more accurately and timely distinguish the consumption records of credit card theft from the normal credit card swiping records,thus providing technical support for the prevention of credit card fraud and personal financial risk control.This paper attempts to obtain credit card swipe transaction data from Kaggle database,and use SMOTE algorithm to deal with data imbalance.Then,the swipe transaction data samples are divided into training set and test set.The combined classifier is trained by Adaboost algorithm,and the classification effect of the combined classifier is compared with that of the traditional logistic regression classification.Finally,it is concluded that the combined classifier can classify the credit card swipetransaction data.The results show that the combined classifier is better than the traditional logistic regression classifier.