Real time credit card fraud detection using computational intelligence

Abstract

Credit card frauds are criminal offences and they should be stopped. If they are not stopped, cardholders, merchants and banks would be affected. Merchants are the most affected party in a credit card fraud. E-commerce has become essential for today‟s global business. Hence, card-not-present fraud committed through e-commerce is becoming a huge widespread problem. This project focuses on creating a model which makes use of computational intelligence as a technique for real time credit card fraud detection. This model combines supervised and unsupervised methods to utilize the strengths and overcome the weaknesses of individual methods. Experiments show that this hybrid model is accurate and feasible for real time credit card fraud detection. This hybrid model aims to demonstrate and highlight the advantages of having both supervised and unsupervised methods in a real time credit card fraud detection model. In this hybrid model, the supervised method used is a General Regression Neural Network while the unsupervised method used is a Kohonen Self Organizing Map Neural Network.