Adaptive Approach of Credit Card Fraud Detection Using Machine Learning Algorithms

E. L. Khyati Bouchra, Abdellatif Ezzouhairi, Khalid Haddouch

Abstract: With the evolution of the electronic payment systems, fraudsters always find an illegal way to steal people's money. Credit card fraud has become a critical issue for businesses and individuals as all online transactions can be easily done by just entering credit card information’s. For this reason, fraud detection is of utmost importance for all financial institutions. Different techniques and approaches are used to secure online transactions, as well as rule-based fraud detection methods, EMV technology, 3D-secure protocol. However, fraud rates still increasing in card not present transactions. Researchers started using different machine learning methods to detect and prevent frauds in online transactions as well as Logistic Regression (LR), Naïve bayes (NB), Random Forest (RF) and Multilayer perceptron (MLP) algorithms. In this paper, the aim objective is to elaborate a comparative study of credit card fraud detection methods. For this objective, we used Google-Colab as an experimentation platform and studied the performance of each machine learning techniques in term of accuracy, Forecast Error and time of prediction using the European cardholder’s dataset that we have balanced by applying SMOTE method. Finally, basing in this comparison study, we proposed and discussed an adaptive approach for credit card detection system.