A response to Webb and Ting's On the application of ROC analysis to predict classification performance under varying class distributions

被引:50
作者
Fawcett, T
Flach, PA
机构
[1] HP Labs, Palo Alto, CA USA
[2] Univ Bristol, Dept Comp Sci, Bristol BS8 1UB, Avon, England
关键词
classification; classifier evaluation; ROC; class skew;
D O I
10.1007/s10994-005-5256-4
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
In an article in this issue, Webb and Ting criticize ROC analysis for its inability to handle certain changes in class distributions. They imply that the ability of ROC graphs to depict performance in the face of changing class distributions has been overstated. In this editorial response, we describe two general types of domains and argue that Webb and Ting's concerns apply primarily to only one of them. Furthermore, we show that there are interesting real-world domains of the second type, in which ROC analysis may be expected to hold in the face of changing class distributions.
引用
收藏
页码:33 / 38
页数:6
相关论文
共 9 条
[1]  
[Anonymous], 1996, P 2 INT C KNOWLEDGE
[2]   Adaptive fraud detection [J].
Fawcett, T ;
Provost, F .
DATA MINING AND KNOWLEDGE DISCOVERY, 1997, 1 (03) :291-316
[3]  
Fawcett T, 2003, HPL20034 HP LAB
[4]  
FLACH P, 2004, ICML04
[5]   Robust classification for imprecise environments [J].
Provost, F ;
Fawcett, T .
MACHINE LEARNING, 2001, 42 (03) :203-231
[6]  
Provost F., 1997, Proceedings of the Third International Conference on Knowledge Discovery and Data Mining, P43
[7]  
Provost F.J., 1998, P 15 INT C MACH LEAR, P445
[8]  
QUINLAN JR, 1987, MACH LEARN, V1, P1
[9]   On the application of ROC analysis to predict classification performance under varying class distributions [J].
Webb, GI ;
Ting, KM .
MACHINE LEARNING, 2005, 58 (01) :25-32