Statistics and Probability Letters (2003), 64 181-189.
Peter G. Hall1 and Rob J. Hyndman1,2
- Centre for Mathematics and its Applications, Australian National University, Canberra ACT 0200, Australia.
- Department of Econometrics and Business Statistics, Monash University, VIC 3800, Australia.
Abstract: The receiver operating characteristic (ROC) curve is used to describe the performance of a diagnostic test which classifies observations into two groups. We introduce new methods for selecting bandwidths when computing kernel estimates of ROC curves. Our techniques allow for interaction between the distributions of each group of observations and gives substantial improvement in MISE over other proposed methods, especially when the two distributions are very different.
Keywords: bandwidth selection; binary classification; kernel estimator; ROC curve.