Ensemble Techniques for Credit Card Fraud Detection

Abstract

Credit card fraud is a problem that has grown by great danger and has a huge impact on the financial sector. The challenges of credit card fraud are the availability of public data, high imbalance in data, and volatility of the fraud nature. Over the years ensemble learning has gained more importance and proved to give better performance. Here we try to do a comparative study of various ensemble approaches using various learning algorithms on the credit card fraud data and to understand multiple models based on various evaluation and performance metrics using the SMOTE balancing technique. machine learning algorithms presented several standard models which include NB, SVM, and DL. They used a publicly available credit card data set has been used for evaluation using individual (standard) models and hybrid models using AdaBoost and majority voting combination methods. The MCC metric was adopted as a performance measure, as it takes into account the true and false positive and negative predicted outcomes. The best MCC score is 0.823, achieved using majority voting. A perfect MCC score of 1 was achieved using AdaBoost and majority voting methods. To further evaluate the hybrid models, noise from 10% to 30% has been added into the data samples. The majority voting method yielded the best MCC score of 0.942 for 30% noise added to the data set. This shows that the majority voting method offers robust performance in the presence of noise. The use of ensemble techniques is very significant in the prediction of faulty credit card transactions from normal credit card transactions.