Fraud Transaction Detection Approach Using Machine Learning Hybrid Techniques

Abstract

Fraud detection is a crucial task in financial transactions to prevent monetary losses and maintain the integrity of the financial system. Some of the machine learning algorithms that we tested for their efficacy in identifying fraudulent activity include Linear Support Vector Classifier (Linear SVC), Support Vector Classifier with Radial Basis Function Kernel (SVC with RBF Kernel), Logistic Regression, Random Forest, Decision Tree, Naive Bayes, Stacking Classifier (Random Forest + SVM with Logistic Regression), and Voting Classifier (Random Forest + Perceptron Algorithm + Boosting. The performance of these algorithms is evaluated using a publicly available dataset consisting of mobile money transactions. We compare the accuracy, precision, recall, F1-score. Our suggested methods demonstrate the capability to accurately identify fraudulent transactions while keeping the number of false positives reasonably low.