Enhancing the Credit Card Fraud Detection Through Ensemble Techniques

Abstract

Nowadays people prefer to use e-commerce because of easiness, timesaving, convenience, etc. By the increase in e-commerce use, credit card fraud increases. The fraudsters get the benefit of online payments and stealing the card details. Therefore, it is essential to improve the detection methods to overcome with the fraudster’s activity and secure the card transactions. The purpose of this study is to investigate the performance of several individual different classifiers and the combination of classifiers using ensemble methods for credit card fraud detection. The study is organized as initially the three well-known classifiers i.e., Decision Tree, Naïve Bayes and SVM have been applied. Afterwards the ensemble learning module have been applied using the boosting technique with the previously mentioned classification algorithms. The dataset used is open source credit card transaction dataset containing 3075 transactions. The performance of the classification techniques is evaluated based on accuracy, sensitivity, specificity, precision, ROC value and F -measure. The result shows that Boosting with Decision Tree outperforms the other techniques.