When studying credit card fraud, researchers often encounter challenges with the imbalance of fraudulent and genuine transactions in data. To address this, Haiyan Kang and Hao Zhang (2022) introduced the K-fold cross-validation and synthetic minority oversampling technique (KCV-SMOTE) and key feature scanning (KFS). This project investigates the KCV-SMOTE and KFS method and its predictive performance under different proportions of unbalanced data and ratios of signal-to-noise using simulated credit transaction data.
This project served as my final research project during my master’s studies at McMaster University. It was made possible with the support and kindness of Dr. Pratheepa Jeganathan.