Skip to main content
. 2011 Sep 9;27(21):3050–3055. doi: 10.1093/bioinformatics/btr516

Fig. 4.

Fig. 4.

Empirical AUC in the validation sets for different methods [A: M1(β); B: M2(β); C: S(β); D: lasso-regularized logistic regression; E: ada-boosting using stumps with 300 iterations] with high-dimensional covariates. Empty box: AUC with one selected covariate; gray box: AUC with three selected covariates; dark gray box, AUC with 10 selected covariates. (a) Simulation setting I (normal); (b) simulation setting II (normal mixture); (c) simulation setting III (log-normal); (d) simulation setting IV (log-normal mixture).