Credit Card Fraud Detection Using Machine Learning and Behavioural Pattern Analysis

Authors: Samarth Aneja, Mr. Ritesh Kumar Chandel

Abstract: Credit card fraud has increased rapidly with the rise of digital payment ecosystems. Traditional rule-based systems struggle to detect evolving attack patterns. This research presents a hybrid approach combining machine learning models with behavioural pattern analysis to improve fraud detection accuracy. Three ML algorithms—Logistic Regression, Random Forest and XGBoost—were trained on an imbalanced credit card transaction dataset. Behavioural features such as spending velocity, merchant category deviation, and location inconsistency were added to enhance model performance. The models were evaluated using precision, recall, F1-score, and AUC values. Results show that incorporating behavioural patterns significantly improves detection rates compared to pure ML models. This approach provides a scalable, real-time method for financial institutions to reduce fraudulent transactions.