Addressing Credit Fraud Threat: Detected Through Supervised Machine Learning Model

In the modern digital age, the rising risk of credit fraud has posed a great challenge to financial institutions and credit card users. This paper follows the analysis of credit card fraud detection notebooks on Kaggle, aiming to improve the accuracy of the automated monitoring system model and the efficiency of institutional operations. Based on realistic credit card transaction records of European cardholders in 2013, a data-driven fraud detection model is trained in the form of a confusion matrix using multiple algorithms as well as various oversampling techniques. The algorithmic models using Random Forest classifier and SMOTE techniques show the best performance based on the assessment criteria of accuracy, specificity and fraud detection rate. From a business application standpoint, in addition to ongoing system maintenance and improvement, credit institutions can work on offering personalized risk monitoring services, negotiating for multi-platform data-sharing cooperation and promoting user education to prevent future fraudulent activities.