Credit Card Fraud Detection Using Machine Learning

Authors: Vinod V Kulkarni, Arghajeet Gupta, Akanksha Kumari Sinha, Farhan Sharieff, Mohan R B

Abstract: Among numerous emerging challenges in the digitized financial ecosystem is credit card fraud. However, traditional rule-based fraud detection systems have rendered fraud detection inadequate in a constantly evolving arena, and the false positives and false negatives are increasing alarmingly. This study implements an accurate and real-time credit card fraud detection using a machine learning- based framework. The system will analyze transaction patterns and classify fraudulent activities using the following multiple classification algorithms: Logistic Regression, Decision Trees, Random Forest, and Gradient Boosting. Procedures concerning preprocessing of the dataset include feature scaling and handling class imbalance through the Synthetic Minority Over-sampling Technique (SMOTE) methodology, followed by dimensionality reduction through PCA, all intended to improve computational efficiency. Results of experiments indicate that ensemble models, and especially Random Forest and XGBoost, produce superior performance with regard to precision, recall, and AUC- ROC scores when compared to baseline models. Results confirm the potential of machine learning in detecting rare fraudulent transactions, as well as scalable solutions for deployment into financial institutions. Enhanced transactional security and reduced losses associated with fraud could be achieved through data-driven predictive modeling.