Ensuring Safe Transactions: Credit Card Fraud Detection with Machine Learning and User-Friendly Interface

Digital transaction growth has increased credit card fraud risks, necessitating effective detection systems for financial security. This research introduces a novel approach combining multiple machine learning models with an intuitive web interface for real-time fraud detection. The hybrid methodology research work uniquely balances accuracy and reduces false positives through feature selection and SMOTE for class imbalance correction. Systematic evaluation of Random Forest, XGBoost, and AdaBoost on a highly imbalanced dataset shows Random Forest achieving 96.83% accuracy with high recall, making it reliable for detecting rare fraudulent transactions. Unlike previous research, this work prioritizes user accessibility via a web-based inter- face, connecting robust backend fraud detection with seamless frontend experience. Results confirm that integrating ensemble learning, class balancing, and real-time feedback improves both detection accuracy and usability, establishing groundwork for future secure digital transaction advancements.