Anomaly and Fraud Detection in Credit Card Transactions Using the ARIMA Model

TL;DR

This paper proposes using the ARIMA time series model for unsupervised credit card fraud detection, demonstrating superior performance over traditional anomaly detection methods on real datasets.

Contribution

It introduces an ARIMA-based approach for fraud detection in unbalanced datasets, outperforming several benchmark anomaly detection techniques.

Findings

ARIMA outperforms K-Means, Box-Plot, LOF, and Isolation Forest in detecting credit card fraud.

The model effectively captures normal spending behavior to identify deviations.

Results show improved detection accuracy over benchmark methods.

Abstract

This paper addresses the problem of unsupervised approach of credit card fraud detection in unbalanced dataset using the ARIMA model. The ARIMA model is fitted on the regular spending behaviour of the customer and is used to detect fraud if some deviations or discrepancies appear. Our model is applied to credit card datasets and is compared to 4 anomaly detection approaches such as K-Means, Box-Plot, Local Outlier Factor and Isolation Forest. The results show that the ARIMA model presents a better detecting power than the benchmark models.