Ensembled-based Credit Card Fraud Detection in Online Transactions

Abstract

Fraud detection systems (FDS) are an important factor in e-commerce trading as online fraud grows and spreads quickly. The incidence of fraud involving online payments by credit cards has recently increased considerably, forcing all banks and e-commerce companies to carry out large-scale operational logs to create automatic methods to detect fraud. The controlled binary classification systems that are appropriately trained from pre-determined sampled data sets appear to be one of the most promising techniques for the detection of unlawful transactions, separating fraudulent and non-fraudulent cases. So the problems related to fraud detection are also classified as a novel challenge that is an assessment based on several advanced algorithms, such as random woodlands, Naive Bays (NB), and generative opponent networks. The problem is a classification problem. In this article, random forest and generative adversary networks are developed to further increase the accuracy of the detection rate of fraud via credit cards. Extensive experiments have shown that the technique offered exceeds current techniques.