Vol. 11 No. 12 (2019): Emerging Trends in Machine Intelligence and Big Data - 1112
Articles

MACHINE LEARNING AND BIG DATA ANALYTICS FOR FRAUD DETECTION SYSTEMS IN THE UNITED STATES FINTECH INDUSTRY

Published 2019-02-04

How to Cite

Saxena, A. K., & Vafin, A. (2019). MACHINE LEARNING AND BIG DATA ANALYTICS FOR FRAUD DETECTION SYSTEMS IN THE UNITED STATES FINTECH INDUSTRY. Emerging Trends in Machine Intelligence and Big Data, 11(12), 1–11. Retrieved from http://orientreview.com/index.php/etmibd-journal/article/view/46

Abstract

The Financial Technology (FinTech) sector in the United States has witnessed rapid growth and innovation, prompting significant changes in the delivery and management of financial services. Alongside these advancements, financial fraud has become increasingly sophisticated, posing challenges to consumer trust and economic stability. This paper investigates the use of machine learning (ML) algorithms and big data analytics for preventing financial fraud in the FinTech environment of the United States. The study evaluates the performance of several machine learning models, including Decision Trees, Support Vector Machines, Random Forests, Neural Networks, and a customized anomaly detection model, in fraud detection. It assesses these models based on metrics such as ROC/AUC scores, True Positives, and False Positive Rates, examining their ability to discern fraudulent transactions from legitimate activities. The Proposed Model outperforms Decision Trees, Support Vector Machines, Random Forests, and Neural Networks with the highest ROC/AUC score of 0.98, despite a varied performance across true positives, false positives, true positive rate, and false positive rate. The study also highlights the role of big data in enhancing fraud detection capabilities, enabling the processing and analysis of large transactional datasets to uncover fraudulent patterns. The research argued that there are challenges, including the lack of universally effective models and the scarcity of comprehensive, publicly available datasets. It advocates for an open exchange of data and insights between financial entities and researchers to foster innovation and improve fraud detection systems. The findings of this study suggest that While machine learning has considerable potential in fraud detection, there is an urgent need for models that adapt dynamically to changing fraud patterns. This paper adds to the area by providing tactical paths for future research and calls for expanded engagement to strengthen the FinTech sector's defenses against an increasing number of financial fraudulent activities.