Credit Card Fraud Detection: Use of Ranking-Based Hybrid Classifier (RBHC) to Improve Forecast Accuracy

Abstract

Credit Card Fraud (CCF) is a major issue in banking and other financial facilities. Now-a-days the usage of
credit cards is frequent in the society. But it is understandable that the number of CCF cases is constantly
increasing in spite of the existing prediction systems. So it is obligatory to find a Credit Card Fraud Detection
(CCFD) System with the highest prediction Accuracy. Different CCFD systems were presented. But these
systems did not give an efficient approach to deal the Class Imbalance problem. In this research, Machine
Learning (ML) algorithms are used to built model for detecting thefrauds in credit card transactions. The
Dataset,used in this research is taken from European cardholders, which consists of 284807 transactions.
One of the problems faced in the existing system was Class Imbalance and this problem is overcome by using
the technique of under sampling in this research. Standard models are first used. Then a novel Rank Based
Hybrid Classifier (RBHC) is proposed to improve the accuracy produced by the standard models. Rank Based
Hybrid Classifier (RBHC) is built after the data set is trained by various Machine Learning algorithms and
ensemble then to predict the final output. The final output on a prediction is taken by majority vote which is
given by the Majority voting strategy. The experimental results indicate that RBHC achieved good accuracy
score of 99.2% in CCFD.