Home / Current Issue / Paper 1705531
Analysis and Comparison of Fraud Detection on Credit Card Transactions Using Machine Learning Algorithms
Subject area: Science,Engineering and Technology · Area of research: Machine Learning
Abstract
Financial organizations and customers are both very concerned about fraud using credit cards as the use of digital payment methods keeps growing. Strong and effective credit card fraud detection systems are essential given the prevalence of complex fraud schemes and the rising number of online transactions. According to transaction statistics, there are more instances of credit card fraud each year. Researchers are thus actively looking on cutting-edge techniques to identify and shut down these fraudulent enterprises. In an effort to safeguard financial institutions and protect consumers from possible losses, they are focused on utilizing cutting- Techniques for better credit card fraud detection and prevention. The purpose of this study is to focus on machine learning approaches. Logistic regression, K-nearest neighbor, and Naive Bayes with stacked model are the techniques used. The algorithms' output is based on f1 score, recall, accuracy, and precision. As well, the ROC curve is plotted. The evaluation of three credit card fraud detection models revealed distinct strengths and weaknesses. While the Naive Bayes model exhibited the highest overall accuracy (99.7%) and F1 score (37.5%), prioritizing precision and recall led to nuanced considerations. Notably, the introduction of an ensemble learning model, stacking Logistic Regression, K-Nearest Neighbor, and Naive Bayes, significantly boosted overall accuracy to an impressive 98.58%, showcasing the potential for enhanced performance through model combination.
Keywords
Logistic Regression, Na?ve Bayes, K-Nearest Neighbor, Fraud Detection, Stacked Generalization
How to cite this paper
@article{1705531,
author = {Ahmad Umar Barmo, Ahmad Haruna, Yusuf Umar Wali, Konika Abid},
title = {Analysis and Comparison of Fraud Detection on Credit Card Transactions Using Machine Learning Algorithms},
journal = {Iconic Research And Engineering Journals},
year = {2024},
volume = {7},
number = {8},
pages = {293-299},
issn = {2456-8880},
url = {https://www.irejournals.com/formatedpaper/1705531.pdf},
abstract = {Financial organizations and customers are both very concerned about fraud using credit cards as the use of digital payment methods keeps growing. Strong and effective credit card fraud detection systems are essential given the prevalence of complex fraud schemes and the rising number of online transactions. According to transaction statistics, there are more instances of credit card fraud each year. Researchers are thus actively looking on cutting-edge techniques to identify and shut down these fraudulent enterprises. In an effort to safeguard financial institutions and protect consumers from possible losses, they are focused on utilizing cutting- Techniques for better credit card fraud detection and prevention. The purpose of this study is to focus on machine learning approaches. Logistic regression, K-nearest neighbor, and Naive Bayes with stacked model are the techniques used. The algorithms' output is based on f1 score, recall, accuracy, and precision. As well, the ROC curve is plotted. The evaluation of three credit card fraud detection models revealed distinct strengths and weaknesses. While the Naive Bayes model exhibited the highest overall accuracy (99.7%) and F1 score (37.5%), prioritizing precision and recall led to nuanced considerations. Notably, the introduction of an ensemble learning model, stacking Logistic Regression, K-Nearest Neighbor, and Naive Bayes, significantly boosted overall accuracy to an impressive 98.58%, showcasing the potential for enhanced performance through model combination.},
keywords = {Logistic Regression, Na?ve Bayes, K-Nearest Neighbor, Fraud Detection, Stacked Generalization},
month = {February},
}