Detection of Credit Card Fraud Utilizing Transaction History Using Machine Learning

Abstract

Over the past several years, fraudulent credit card transactions have surged, posing significant challenges to financial institutions and consumers alike. In 2021, an annual study revealed that over 50% of Americans experienced credit or debit card fraud, with approximately 127 million individuals falling victim at least once. Traditional fraud detection techniques are often inadequate due to their reliance on manual processes and rule-based systems, which are both time-consuming and prone to errors. This paper explores the application of machine learning (ML) algorithms to enhance the accuracy and efficiency of credit card fraud detection. Six supervised ML algorithms—Naïve Bayes, Support Vector Machine (SVM), Random Forest, K-Nearest Neighbors (KNN), Logistic Regression, and XGBoost—are evaluated using transaction history data to classify fraudulent and non-fraudulent activities. The study employs a comprehensive methodology involving data preprocessing, feature engineering, and model training, followed by performance evaluation based on accuracy, precision, recall, and F1-score. Results indicate that the SVM model outperforms other algorithms, achieving the highest accuracy in fraud classification. The findings underscore the potential of ML in automating and improving fraud detection systems, thereby mitigating financial losses and enhancing trust in financial services.