Credit Card Fraud Detection Using Machine Learning Techniques

Credit card fraud poses a persistent challenge in today's e-commerce market, with a significant surge in fraudulent activities observed over recent years. This surge has resulted in financial losses for numerous organizations, companies, and government agencies. As digital payments continue to gain popularity, it is expected that the problem of credit card fraud will only intensify in the future. To address this issue, this study concentrates on the early identification of fraudulent online credit card transactions, utilizing advanced machine learning techniques. However, detecting credit card fraud proves to be a complex undertaking due to the highly imbalanced nature of the dataset. In other words, the frequency of genuine cases far exceeds that of fraudulent cases. To tackle this problem, predictive models such as logistic regression, random forest, k-nearest neighbors, and XGBoost, in conjunction with various resampling techniques, have been employed on the imbalanced dataset to ascertain whether a transaction is fraudulent or genuine. The experimental findings demonstrated that the ensemble method of max voting, combined with a hybrid resampling approach known as the synthetic minority oversampling technique (SMOTE), yielded the most favorable outcomes.