CPRS-Inc. Grant: Analysis of Duplicate/ Erroneous Transactions & Related Identification Techniques
COMPETITIVE GRANT AWARD WINNER By their nature, competitive grants are challenging to receive, as faculty must make a case for why their research should be funded over other projects. This award recognizes top-quality research completed as part of competitive grants from organizations like the National Science Foundation, National Institute of Health, Department of Homeland Security, The Institute of Internal Auditors, and more.
Aaron French
This research project, supported by a $53,000 grant in collaboration with CPRS, aimed to enhance duplicate payment detection using machine learning techniques. CPRS is a leader in AP audit services specializing in recovering duplicate payments and other overpayments by analyzing data beyond typical ERP system checks. Our project developed an advanced tool to complement CPRS’s existing detection processes, introducing machine learning models to identify
payments meeting a minimum duplication threshold with assigned probabilities. This innovation enables faster auditing and more efficient payment recovery. Through comprehensive data analysis, we found that duplicate payments occurred in less than one percent of all vendors but identified several vendors with multiple duplicate payment instances. These findings resulted in recommendations to improve business processes and targeting recommendations for
business to address recurring issues with problem vendors. Our research advances automated auditing capabilities, supporting CPRS’s mission to maximize payment recovery and strengthen vendor accountability. Based on success of the project, the grant was renewed for spring 2025 contributing to a total of $106,000 for the 2024-2025 academic year with new business problems to solve.
Organization Awarding Grant: https://cprs-inc.com/ap-audit-services/
Complex systems and messy data require significant transformation for effective ML use. Data analysis is more than outcomes; it requires understanding the story of data. ML tools enabled the identification of duplicate payment well below standard thresholds. This success resulted in a contract renewal to address other business problems.
19
Made with FlippingBook - Online catalogs