Credit Card Fraud Detection Based on ANN and XGBoost

Junfeng Shi

2024

Abstract

Due to the rapid expansion of electronic payment systems, the number of people using credit cards has grown significantly. However, as credit cards provide massive convenience for the public, the number of related fraud cases subsequently increased, causing significant losses for the public and the credit card issuing banks. The traditional method of fraud detection requires an extended manual analysis, which becomes almost impossible given the massive popularity of credit cards. Hence, establishing an effective credit card fraud detection (CCFD) system is imperative to reverse the situation. ANN and XGBoost are two powerful algorithms for classification problems. Their performances on balanced data sets have already been recognized, while their performances on imbalanced data sets remain unknown. To discover whether these two algorithms are suitable for CCFD, this paper applies ANN and XGBoost to the binary classification problem of CCFD and analyses their performance. The result shows that the accuracy rate of both ANN and XGBoost is as high as 99.96%. However, the f1 score of XGBoost on the minority class is higher than ANN's, indicating that XGBoost can identify the minority class more efficiently. Therefore, XGBoost is a better option for credit card detection than ANN.

Paper Citation

in Harvard Style

Shi J. (2024). Credit Card Fraud Detection Based on ANN and XGBoost. In Proceedings of the 1st International Conference on E-commerce and Artificial Intelligence - Volume 1: ECAI; ISBN 978-989-758-726-9, SciTePress, pages 195-199. DOI: 10.5220/0013212700004568

in Bibtex Style

@conference{ecai24,
author={Junfeng Shi},
title={Credit Card Fraud Detection Based on ANN and XGBoost},
booktitle={Proceedings of the 1st International Conference on E-commerce and Artificial Intelligence - Volume 1: ECAI},
year={2024},
pages={195-199},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0013212700004568},
isbn={978-989-758-726-9},
}

in EndNote Style

TY - CONF

JO - Proceedings of the 1st International Conference on E-commerce and Artificial Intelligence - Volume 1: ECAI
TI - Credit Card Fraud Detection Based on ANN and XGBoost
SN - 978-989-758-726-9
AU - Shi J.
PY - 2024
SP - 195
EP - 199
DO - 10.5220/0013212700004568
PB - SciTePress