Automated Techniquesfor Detectionof Procurement Fraud

This document describes automated techniques for detecting procurement fraud in large datasets. It provides a case study and methodology based on a real fraud that occurred in the Wake County Public School System. The methodology involves extracting relevant payment data, performing initial analyses to identify outliers, and then drilling down on specific vendors identified as outliers. The analyses include histograms, Benford's Law testing, payment trends by month, stratifying payments by purchase order amount, and identifying consecutively numbered or holiday-dated invoices. The goal is to efficiently filter large datasets to identify anomalies worth further investigation that may indicate fraud. Sample data and scripts are provided to demonstrate the techniques.