USING ADVANCED ANALYTICS IN FRAUD DETECTION AND FINANCIAL CRIME PREVENTION IN FINANCIAL INSTITUTIONS ACROSS THE UNITED STATES

Main Article Content

Muleka Christelle Masudi

Abstract

: Banks and financial institutions in the U.S. are facing more and more challenges in fighting fraud and financial crime. Crimes like identity theft, money laundering, account takeovers, and payment scams are growing, especially as more people use mobile banking, peer-to-peer apps, and instant payments. Traditional fraud systems based on fixed rules can’t keep up with modern scams, which now include fake identities, bot attacks, and complicated laundering schemes (Consumer Financial Protection Bureau [CFPB], 2023; KPMG, 2025). This study looks at how advanced analytics, including machine learning (ML), anomaly detection, and natural language processing (NLP) can help stop fraud more effectively. We used transaction data from five U.S. banks (from 2022 to 2024) and trained smart models using customer behavior, time patterns, and location data.


The results were impressive: Fraud detection accuracy went up from 74% to 93%, and False alarms dropped by 48%


New types of fraud (missed by older systems) were caught 70% of the time using unsupervised models like clustering and autoencoders. There was also a strong match between the model’s risk scores and real fraud cases (correlation of r = 0.88, p < .001) (Federal Reserve, 2024). NLP tools were also successful, reaching an F1-score of 0.89, in identifying issues in transaction notes and documents like fake companies, unclear ownership, or risky countries (FinCEN, 2023). These findings show that using analytics makes fraud detection faster, smarter, and more flexible. It helps banks catch fraud in real time, follow government rules (like the Bank Secrecy Act and AML regulations), and improve fraud investigation and reporting. This paper recommends that banks use these analytics tools throughout the entire fraud prevention process, from live monitoring to post-fraud reviews and compliance reporting.

Downloads

Download data is not yet available.

Article Details

Section
Articles