Benford-Newcomb Subsequences for Fraud Detection

TL;DR

This paper introduces a novel approach using Benford-Newcomb subsequences to detect data fraud by comparing leading digit frequencies to natural number probabilities, establishing interval criteria for further investigation.

Contribution

It proposes a new method leveraging Benford-Newcomb subsequences and natural number probabilities for improved fraud detection.

Findings

Interval criteria effectively identify suspicious data sets

Method outperforms traditional Benford's law tests in certain scenarios

Provides a new statistical tool for forensic data analysis

Abstract

Benford's law is frequently used to evaluate the likihood that data is misrepresentative. Typically statistical tests measure the likihood. Another method of employing Benford's law is to compare the frequency of leading digits to the probabilities of leading digits over a subset of the natural numbers. This paper proposes using the probabilities of leading digits from uniform, natural numbers to establish interval criteria for when to look more closely into the possibility of misrepresentative data.