Classification by pairwise coupling
Hastie, Trevor ; Tibshirani, Robert
Ann. Statist., Tome 26 (1998) no. 3, p. 451-471 / Harvested from Project Euclid
We discuss a strategy for polychotomous classification that involves coupling the estimating class probabilities for each pair of classes, and estimates together. The coupling model is similar to the Bradley-Terry method for paired comparisons. We study the nature of the class probability estimates that arise, and examine the performance of the procedure in real and simulated data sets. Classifiers used include linear discriminants, nearest neighbors, adaptive nonlinear methods and the support vector machine.
Publié le : 1998-04-14
Classification:  Pairwise,  Bradley-Terry model,  62H30,  68T10,  62J15
@article{1028144844,
     author = {Hastie, Trevor and Tibshirani, Robert},
     title = {Classification by pairwise coupling},
     journal = {Ann. Statist.},
     volume = {26},
     number = {3},
     year = {1998},
     pages = { 451-471},
     language = {en},
     url = {http://dml.mathdoc.fr/item/1028144844}
}
Hastie, Trevor; Tibshirani, Robert. Classification by pairwise coupling. Ann. Statist., Tome 26 (1998) no. 3, pp.  451-471. http://gdmltest.u-ga.fr/item/1028144844/