Data Mining of Operations with Card Accounts of Bank Clients

Abstract

The article is devoted to the expediency of using the data mining and the construction of the neural network for the evaluation of transactions with card accounts for detecting attempts of frauds. The authors proposed a scheme for customer interaction with the bank when transaction is performing with the payment cards. The process is carried out using the verification module with data mining. The article was built a neural network with using software “Statistica”. The authors selected a data set that contains amounts of transaction, time intervals, fraud identifiers. As a result, it was got a multilayer perceptron with nine inputs, five hidden neurons and two outputs that can be used to predict an attempt at fraud with card accounts of bank clients.