Credit Card Fraud Detection

Mayowa Timothy Adesina *and Luke Howe

Data Analytics Department, College of Business, Kansas State University, KS, USA.
 
Research Article
International Journal of Science and Research Archive, 2024, 12(02), 2072–2080.
Article DOI: 10.30574/ijsra.2024.12.2.1430
Publication history: 
Received on 05 July 2024; revised on 12 August 2024; accepted on 14 August 2024
 
Abstract: 
Credit card fraud poses a significant threat globally, impacting both individuals and businesses. Leveraging machine learning techniques for fraud detection is a powerful strategy, capitalizing on the ability of algorithms to analyze vast amounts of historical transaction data. These models learn intricate patterns and anomalies indicative of fraudulent behavior, enabling them to discern subtle deviations from normal spending patterns. By considering factors like transaction frequency, location, and user behavior, machine learning systems can dynamically adapt and evolve, staying ahead of sophisticated fraud tactics. This approach not only enhances the accuracy of fraud detection but also allows for real-time monitoring, enabling swift intervention to prevent unauthorized transactions and safeguard financial assets, ultimately fortifying the resilience of credit card systems in the face of evolving security threats.
In this paper, we introduce a machine learning credit card fraud detection model leveraging the Random Forest algorithm, a well-established ensemble learning technique that amalgamates multiple decision trees for enhanced predictive accuracy. The study specifically addresses the inherent challenge of class imbalance in credit card fraud datasets by comparing the performance of Random Forest under various fine-tuning methods. These methods include random oversampling, class weights adjustment, as well as the utilization of both smoke and Tomek links (a method for under sampling) and smoke oversampling.
We evaluate our model on a real-world credit card fraud dataset from Kaggle, and show that our model achieves high accuracy, precision, recall, and F1-score, while reducing the false positive rate and the false negative rate. We also discuss the advantages and limitations of our model, and suggest some possible directions for future work.
This study adds to the expanding realm of machine learning applications in credit card fraud detection, with implications reaching beyond the financial sector into diverse industries. The objective is to optimize the model's efficacy in detecting fraudulent transactions by mitigating the challenges posed by imbalanced data.
 
Keywords: 
Machine Learning; Artificial Intelligence; Credit Card; Fraud; Random Forest; Imbalanced Dataset; Financial Industry.
 
Full text article in PDF: 
 
Copyright information: 

Copyright © 2024 Author(s) retain the copyright of this article. This article is published under the terms of the Creative Commons Attribution Liscense 4.0