ML Anthology

On the Application of ROC Analysis to Predict Classification Performance Under Varying Class Distributions

Geoffrey I. Webb, Kai Ming Ting
MLJ 2005 pp. 25-32
doi:10.1007/S10994-005-4257-7 /mlj/2005/webb2005mlj-application/

Abstract

We counsel caution in the application of ROC analysis for prediction of classifier performance under varying class distributions. We argue that it is not reasonable to expect ROC analysis to provide accurate prediction of model performance under varying distributions if the classes contain causally relevant subclasses whose frequencies may vary at different rates or if there are attributes upon which the classes are causally dependent.

PDF MLJ Semantic Scholar

Cite

Text

Webb and Ting. "On the Application of ROC Analysis to Predict Classification Performance Under Varying Class Distributions." Machine Learning, 2005. doi:10.1007/S10994-005-4257-7

Markdown

[Webb and Ting. "On the Application of ROC Analysis to Predict Classification Performance Under Varying Class Distributions." Machine Learning, 2005.](https://mlanthology.org/mlj/2005/webb2005mlj-application/) doi:10.1007/S10994-005-4257-7

BibTeX

@article{webb2005mlj-application,
  title     = {{On the Application of ROC Analysis to Predict Classification Performance Under Varying Class Distributions}},
  author    = {Webb, Geoffrey I. and Ting, Kai Ming},
  journal   = {Machine Learning},
  year      = {2005},
  pages     = {25-32},
  doi       = {10.1007/S10994-005-4257-7},
  volume    = {58},
  url       = {https://mlanthology.org/mlj/2005/webb2005mlj-application/}
}