Best subsets of certain size along the regularization path of the logistic regression classifier with average marker intensities as features.

Markers in the list have been sorted according to the absolute value of their corresponding coefficient such that the most important feature is always on top. CV err. indicates the 10-fold CV classification error when using regularization parameter corresponding to the norm of the coefficient vector as given in . The lowest CV error () is attained with the 17 markers shown in Figure 6.