Credit Card Security Enhancement Through Integrated Machine Learning Techniques

Once upon a time, people exclusively used cash for transactions. The advent of the digital era ushered in new technologies, including the introduction of online transactions facilitated by credit cards. However, the increased reliance on credit cards also gave rise to a global issue-credit card fraud. Over time, this fraudulent activity has surged, posing a significant threat. These illicit activities not only compromise the financial security of individuals but also pose a risk to businesses and financial institutions. Credit card fraud represents a small fraction of total transactions, creating an imbalanced dataset that challenges effective learning for models. Initially, systems with predefined rules were utilized to identify transactions possessing higher risks and trigger alerts. However, these systems lacked adaptability and were susceptible to attacks. Traditional credit card fraud detection systems struggle to combat fraudsters who employ sophisticated techniques such as data breaches, phishing, and scams to compromise credit card information for illicit transactions. Our research utilized a varied selection of machine learning techniques, such as XGBoost, KNN, LGBM, Logistic Regression, Random Forest, and Decision Trees, to construct a comprehensive system aimed at identifying fraudulent behaviour. The design incorporates a stacking method to identify fraudulent activities more effectively and efficiently. This approach not only works for credit card fraud but also contributes towards protecting customers' financial interests.