THESIS AN OVERVIEW OF CREDIT CARD FRAUD DETECTION LEARNING TECHNIQUES

Abstract

Credit card fraudsters are becoming more creative, altering their behaviors, and finding new ways to trick computer systems. Card fraud has become a major national and global threat to e-commerce causing losses of great amounts of money. Immediate attention needs to be directed towards improving existing techniques, or creating new methods for pinpointing fraudulent transactions. Supervised classification algorithms have proven to be accurate measures for predicting illegal transaction with more than 90% accuracy. This work reviews existing techniques and compares their reliability by examining their accuracy and speed on their application to three deferent data sets.