Browse > Article
http://dx.doi.org/10.5351/CKSS.2004.11.1.021

ROC and Cost Graphs for General Cost Matrix Where Correct Classifications Incur Non-zero Costs  

Kim, Ji-Hyun (Department of Statistics, Soongsil University)
Publication Information
Communications for Statistical Applications and Methods / v.11, no.1, 2004 , pp. 21-30 More about this Journal
Abstract
Often the accuracy is not adequate as a performance measure of classifiers when costs are different for different prediction errors. ROC and cost graphs can be used in such case to compare and identify cost-sensitive classifiers. We extend ROC and cost graphs so that they can be used when more general cost matrix is given, where not only misclassifications but correct classifications also incur penalties.
Keywords
cost-sensitive learning; cost matrix; iso-performance line; CART;
Citations & Related Records
연도 인용수 순위
  • Reference
1 Signal Detection Theory and ROC Analysis /
[ Egan,J.P. ] / Series in Cognition and Perception
2 The fouondations of cost-sensitive learning /
[ Elkan,C. ] / Proceedings of the Seventeenth International Joint Conference on Artificial Intelligence
3 Analysis and visualization of classifier performance:Comparison under imprecise class and cost distributions /
[ Provost,F.;Fawcett,T. ] / Proceedings of the Third International Conference on Knowledge Discovery and Data Mining
4 Explicitly representing expected cost: An alternative to ROC representation /
[ Drummond,C.;Holte,R. ] / Technical Report, School of Information Technology and Engineering, University of Ottawa
5 /
[ Blake,C.L.;Merz,C.J. ] / UCI Repository of machine iearning databases
6 An introduction to recursive partitioning using the RPART routines /
[ Therneau,T.M.;Atkinson,E.J. ] / Technical Report, Mayo Foundation
7 /
[ Breiman,L.;Friedman,J.H.;Olshen,J.A.;Stone,C.J. ] / Classification and Regression Trees