Figure 2. Repeated cross-validation of class labels assigned using receptor status in breast cancer and normal breast tissue samples.

A) One-against-one feature selection and classification. B) One-against-all feature selection and classification. LIMMA, linear models for microarray; OVL, overlap of locally adaptive kernel densities; WSV, weight of support vectors; BAG, bagging; J48, C4.5 decision tree; KNN, k-nearest neighbors, NBA, naive Bayes; RFO, random forests; SVM, support vector machine.