Validation of Two-Sample Bootstrap in ROC Analysis on Large Datasets Using AURC

Published

October 11, 2010

Author(s)

Jin Chu Wu, Alvin F. Martin, Raghu N. Kacker

Abstract

Sampling variability can result in uncertainties of measures. The nonparametric two-sample bootstrap method has been used to compute uncertainties of measures in receiver operating characteristic (ROC) analysis on large datasets, such as the standard error (SE) of the equal error rate in biometrics, the SE of a detection cost function in speaker recognition evaluation, etc. It is hard to calculate uncertainties of these statistics of interest without using bootstrap methods. The SE of the area under ROC curve (AURC) can be computed analytically using the Mann-Whitney statistic. It can also be calculated using the nonparametric two-sample bootstrap method. The analytical result could be treated as a ground truth. The relative errors of bootstrap-method results with respect to the analytical-method results using different matching algorithms were examined. It turned out that they were quite small. Hence, this validates the nonparametric two-sample bootstrap method applied in ROC analysis on large datasets.

Citation

NIST Interagency/Internal Report (NISTIR) - 7733

Report Number

7733

NIST Pub Series

NIST Interagency/Internal Report (NISTIR)

Pub Type

NIST Pubs

Download Paper

Local Download

Keywords

ROC analysis, bootstrap, area under ROC curve, uncertainty, standard error, biometrics, speaker recognition

Citation

, J. , Martin, A. and Kacker, R. (2010), Validation of Two-Sample Bootstrap in ROC Analysis on Large Datasets Using AURC, NIST Interagency/Internal Report (NISTIR), National Institute of Standards and Technology, Gaithersburg, MD, [online], https://tsapps.nist.gov/publication/get_pdf.cfm?pub_id=906231 (Accessed October 4, 2024)