STRENGTHENING AMERICAN LEADERSHIP IN DIGITAL FINANCIAL TECHNOLOGY THROUGH MACHINE LEARNING

Abstract

The last several years have seen a notable improvement in financial inclusion in America.  Bank accounts have become more common among Indians in recent years. The growing risk of credit card fraud resulting from the use of digital financial transactions calls for the development of clever investigation techniques. Therefore, in order to propose a machine learning-based fraud detection system, this study will use the American Express Credit Card Fraud Detection (CCFD) dataset. The stages involved in data preparation include managing missing values, normalization, noise reduction, and SMOTE class balance. To increase model efficiency, feature selection is done using statistical approaches. The XGBoost algorithm's High Accuracy and Scalability are employed for categorization. The suggested model performs better than a number of baseline models in terms of accuracy (98.7%), precision (96.4%), recall (95.8%), and F1-score (96.1%), according to experimental study results. These findings demonstrate that the model can most successfully identify fraudulent transactions with a low rate of false positives. The study adds to strengthening digital financial security and supports the development of FinTech leadership in America.