Enhanced credit card fraud detection based on SVM-recursive feature elimination and hyper-parameters optimization

Abstract

With the growth of online shopping, Credit Card Fraud (CCF) comes out as a serious menace. For this end, the automatic and real-time fraud detection field calls for several studies. The recent ones use many Machine Learning (ML) methods and techniques due to their beneficial characteristics to build a good fitting model to catch fraudulent transactions. The purpose of this paper is to develop a new model based on a hybrid approach for Credit Card Fraud Detection (CCFD). The proposed model, compared to previous studies, shows its strong ability to identify fraud transactions. Precisely, the robustness of our model is built by combining the strength of three sub-methods; the Recursive Feature Elimination (RFE) for selecting the most useful predictive features, the GridSearchCV for Hyper-Parameters Optimization (HPO) and the Synthetic Minority Oversampling (SMOTE) to overcome the imbalanced data problem. The experimentations of our model, on many real-world data sets, gives the best results in terms of efficiency and effectiveness.