Detect deception on banking credit card payment system by machine learning classifiers
Md Babul Islam, Khandaker Sajidul Islam, Md Helal Khan, Abdullah MMA Al Omari, Swarna Hasibunnahar
Abstract
Credit cards are now being targeted by fraudsters in a variety of ways. Despite the fact that it is never pleasant, this is something that happens to someone in our everyday life. When cardholders disclose their credit card information to others, this happens. In this work, we have constructed a few machines learning (ML) models using anonymous credit card transaction data. The issue in detecting fraud is that it occurs far less frequently than legal transactions. The purpose of this research is to accurately predict fraud transactions. To detect fraud from a vast unbalanced dataset, we used nine different classifiers (Ridge Classifiers, Stochastic Gradient Descent (SGD), Linear discriminant analysis (LDA), Random Forest, Naive Bayes, Support Vector Machine (SVM), Decision Tree, Logistic Regression, and k-Nearest Neighbors (k-NN)). In addition, various classifiers were compared to ROC binary classifications. We have shown which classifiers has the best accuracy.