Explicitly representing expected cost: an alternative to ROC representation

作者: Chris Drummond , Robert C. Holte

DOI: 10.1145/347090.347126

关键词: Class (set theory)Point (geometry)Pattern recognitionDuality (mathematics)Artificial intelligenceLine (geometry)Computer scienceLine segmentSpace (mathematics)Range (mathematics)Representation (mathematics)

摘要: ABSTRACT This paper proposes an alternative to ROC representation, in which the expected cost of a classi er is represented explicitly. representation maintains many advantages but easier understand. It allows experimenter immediately see range costs and class frequencies where particular best quantitatively how much better it than other ers. demonstrates there point/line duality between two representations. A point space representing becomes line segment spanning full frequencies. produces equivalent operations spaces, allowing most techniques used analysis be readily reproduced space.

参考文章(13)
John Arthur Swets, Ronald M. Pickett, Evaluation of diagnostic systems : methods from signal detection theory Academic Press. ,(1982)
Tom Fawcett, Foster Provost, Adaptive Fraud Detection Data Mining and Knowledge Discovery. ,vol. 1, pp. 291- 316 ,(1997) , 10.1023/A:1009700419189
Foster Provost, R Fawcett, T, Kohavi, The Case against Accuracy Estimation for Comparing Induction Algorithms international conference on machine learning. pp. 445- 453 ,(1998)
Foster Provost, Tom Fawcett, Robust classification systems for imprecise environments national conference on artificial intelligence. pp. 706- 713 ,(1998)
Lee B. Lusted, Introduction to medical decision making Charles C. Thomas. ,(1968)
Miroslav Kubat, Robert C. Holte, Stan Matwin, Machine Learning for the Detection of Oil Spills in Satellite Radar Images Machine Learning. ,vol. 30, pp. 195- 215 ,(1998) , 10.1023/A:1007452223027
Richard A Olshen, Charles J Stone, Leo Breiman, Jerome H Friedman, Classification and regression trees ,(1983)
J. Swets, Measuring the accuracy of diagnostic systems Science. ,vol. 240, pp. 1285- 1293 ,(1988) , 10.1126/SCIENCE.3287615
Pedro Domingos, MetaCost: a general method for making classifiers cost-sensitive knowledge discovery and data mining. pp. 155- 164 ,(1999) , 10.1145/312129.312220