Credit Card Fraud Detection Based on Machine Learning

Citations Over TimeTop 10% of 2019 papers

Abstract

In recent years, the rapid development of e-commerce exposes great vulnerabilities in online transactions for fraudsters to exploit. Credit card transactions take a salient role in nowadays’ online transactions for its obvious advantages including discounts and earning credit card points. So credit card fraudulence has become a target of concern. In order to deal with the situation, credit card fraud detection based on machine learning is been studied recently. Yet, it is difficult to detect fraudulent transactions due to data imbalance (normal and fraudulent transactions), for which Smote algorithm is proposed in order to resolve data imbalance. The assessment of Light Gradient Boosting Machine model which proposed in the paper depends much on datasets collected from clients’ daily transactions. Besides, to prove the new model’s superiority in detecting credit card fraudulence, Light Gradient Boosting Machine model is compared with Random Forest and Gradient Boosting Machine algorithm in the experiment. The results indicate that Light Gradient Boosting Machine model has a good performance. The experiment in credit card fraud detection based on Light Gradient Boosting Machine model achieved a total recall rate of 99% in real dataset and fast feedback, which proves the new model’s efficiency in detecting credit card fraudulence.

Related Papers

• → Semi-Supervised Classification on Credit Card Fraud Detection using AutoEncoders(2021)18 cited
• → Prevention of credit card fraud detection based on HSVM(2016)23 cited
• → Effective Machine Learning Approaches for Credit Card Fraud Detection(2021)9 cited
• → Gradient Boosting Machine: A Survey(2019)25 cited
• → Survey Paper on Credit card Fraud Detection System(2022)