Performance criteria for plastic card fraud detection tools

Authors: Hand, D J¹; Whitrow, C²; Adams, N M³; Juszczak, P²; Weston, D²

Source: Journal of the Operational Research Society, Volume 59, Number 7, 23 July 2008 , pp. 956-962(7)

Buy & download fulltext article:

The full text article is not available.

Abstract:

In predictive data mining, algorithms will be both optimized and compared using a measure of predictive performance. Different measures will yield different results, and it follows that it is crucial to match the measure to the true objectives. In this paper, we explore the desirable characteristics of measures for constructing and evaluating tools for mining plastic card data to detect fraud. We define two measures, one based on minimizing the overall cost to the card company, and the other based on minimizing the amount of fraud given the maximum number of investigations the card company can afford to make. We also describe a plot, analogous to the standard ROC, for displaying the performance trace of an algorithm as the relative costs of the two different kinds of misclassification—classing a fraudulent transaction as legitimate or vice versa—are varied.Journal of the Operational Research Society (2008) 59, 956-962. doi:10.1057/palgrave.jors.2602418 Published online 9 May 2007

Document Type: Research article

Affiliations: 1: 1: 1Department of Mathematics, Imperial College, London, UK 2: 2Institute for Mathematical Sciences, Imperial College, London, UK 2: 2Institute for Mathematical Sciences, Imperial College, London, UK 3: 1Department of Mathematics, Imperial College, London, UK

Publication date: 2008-07-23