Title: A stacking ensemble for credit card fraud detection using SMOTE technique

Authors: Kaithekuzhical Leena Kurien; Ajeet A. Chikkamannur

Addresses: Visvesvaraya Technological University Belagavi, Karnataka, 590018, India; Department of Computer Science and Engineering, R.L. Jalappa Institute of Technology, Doddaballapur, Bengaluru, 561203, India ' Nagarjuna College of Engineering and Technology, Beedaganahalli, Venkatagiri Kote, Post, Devanahalli, Bengaluru, Karnataka, 562110, India

Abstract: In present times, online shopping is on rise as more people stay at home and credit card do the walking resulting in increase in credit card fraud. COVID-19 outbreak pushed digital payments exponentially, making gateway for online frauds even more common to this dependence on digital payment platforms. Empirical evidence on ensemble techniques of machine learning algorithms for fraud prediction has exhibited superior performance in identifying fraud patterns in usage of credit card data through technique of stacking and voting. The ensemble techniques like AdaBoost, gradient boost and XGBoost are applied along with stacking classifier on the credit card dataset. The stacking classifier using heterogeneous classifiers of random forest, K-nearest neighbours and logistic regression to learn the fraud patterns effectively in credit card data. The proposed methods are analysed using F1-score and recall metrics due to skewness of data.

Keywords: ensemble methods; random forest; fraud detection; XGBoost; AdaBoost; anomaly detection.

DOI: 10.1504/IJESMS.2024.141968

International Journal of Engineering Systems Modelling and Simulation, 2024 Vol.15 No.6, pp.284 - 290

Received: 13 Nov 2021
Accepted: 02 Feb 2022
Published online: 04 Oct 2024 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article