Forensic Analytics, 2nd Edition

CHAPTER 6Identifying Anomalous Outliers: Part 1

THE NIGRINI CYCLE TESTS used the amounts (usually dollars) in a single field only. The data profile and the histogram gave us information on the distribution of the amounts. The periodic graph used the monthly totals. The various digit-based tests and the number duplication test all analyzed the digit and number patterns in a single field. In the chapters that follow the analytics tests will analyze one or more fields at once. These more focused tests will give us smaller groups of notable records. The tests that follow are of a more advanced nature.

The tests in this chapter and the following chapter are focused on outliers, which are data points that differ significantly from the norm. In statistical theory outliers could be caused by an apparatus making a measurement error, an error in data transmission, elements outside the population being incorrectly included in the population, a flaw in an assumed theory, and so on. In our domain we are concerned with outliers being a red flag for fraud and/or errors.

The first test is called the summation test and this test identifies abnormally large amounts in the data set. The second test is the largest subsets test. The test uses two fields, one with transaction or balance numbers (such as dollars, inventory counts, vote counts, population counts) and another field to indicate the subset (e.g., vendor number, credit card number, or branch number). The word subset is borrowed from ...