Faculty Digital Archive

Archive@NYU >
Stern School of Business >
CeDER Published Papers >

Please use this identifier to cite or link to this item: http://hdl.handle.net/2451/27808

Title: Pointwise ROC Confidence Bounds: An Empirical Evaluation
Authors: Macskassy, Sofus
Provost, Foster
Rosset, Saharon
Issue Date: 2005
Citation: Proceedings of the ICML 2005 workshop on ROC
Series/Report no.: CeDER-PP-2005-06
Abstract: This paper is about constructing and evaluating pointwise confidence bounds on an ROC curve. We describe four confidencebound methods, two from the medical field and two used previously in machine learning research. We evaluate whether the bounds indeed contain the relevant operating point on the “true” ROC curve with a confidence of 1− . We then evaluate pointwise confidence bounds on the region where the future performance of a model is expected to lie. For evaluation we use a synthetic world representing “binormal” distributions–the classification scores for positive and negative instances are drawn from (separate) normal distributions. For the “true-curve” bounds, all methods are sensitive to how well the distributions are separated, which corresponds directly to the area under the ROC curve. One method produces bounds that are universally too loose, another universally too tight, and the remaining two are close to the desired containment although containment breaks down at the extremes of the ROC curve. As would be expected, all methods fail when used to contain “future” ROC curves. Widening the bounds to account for the increased uncertainty yields identical qualitative results to the “true-curve” evaluation. We conclude by recommending a simple, very efficient method (vertical averaging) for large sample sizes and a more computationally expensive method (kernel estimation) for small sample sizes.
URI: http://hdl.handle.net/2451/27808
Appears in Collections:CeDER Published Papers

Files in This Item:

File Description SizeFormat
CPP-06-05.pdf205.79 kBAdobe PDFView/Open

Items in Faculty Digital Archive are protected by copyright, with all rights reserved, unless otherwise indicated.


The contents of the FDA may be subject to copyright, be offered under a Creative Commons license, or be in the public domain.
Please check items for rights statements. For information about NYU’s copyright policy, see http://www.nyu.edu/footer/copyright-and-fair-use.html 
Valid XHTML 1.0 | CSS