Table 1. Dimension Reduction Classifier Performance Summary.

Method	MCR	AUC	Spec	Sens	No. Components	Data set
pls.lda (full)	0.287	0.635	0.694	0.75	12	Gaucher
pls.rf (full)	0.343	0.75	0.707	0.636	3
pca.lda (full)	0.231	0.992	0.779	0.794	17
pls.lda (trimmed)	0.115	0.823	0.859	0.918	7
pls.rf (trimmed)	0.171	0.919	0.852	0.821	7
pca.lda (trimmed)	0.046	0.992	0.918	0.995	6
pls.lda.lung (full)	0.196	0.889	0.778	0.837	5	Lung cancer
pls.rf.lung (full)	0.225	0.85	0.764	0.794	6
pca.lda.lung (full)	0.2	0.897	0.756	0.85	9
pls.lda (trimmed)	0.199	0.881	0.79	0.819	5
pls.rf (trimmed)	0.232	0.841	0.751	0.794	6
pca.lda (trimmed)	0.217	0.88	0.741	0.83	17
pls.lda.CRC (full)	0.089	0.954	0.853	0.966	5	Colorectal cancer
pls.rf.CRC (full)	0.113	0.951	0.88	0.896	8
pca.lda.CRC (full)	0.089	0.97	0.862	0.959	10
pls.lda (trimmed)	0.089	0.951	0.855	0.963	6
pls.rf (trimmed)	0.119	0.941	0.89	0.877	8
pca.lda (trimmed)	0.11	0.952	0.845	0.933	2
pls.lda (full)	0.26	0.478	0.7	0.784	8	Ovarian cancer
pls.rf (full)	0.286	0.794	0.678	0.722	7
pca.lda (full)	0.315	0.777	0.627	0.757	17
pls.lda (trimmed)	0.159	0.914	0.807	0.811	3
pls.rf (trimmed)	0.191	0.897	0.818	0.81	9
pca.lda (trimmed)	0.157	0.931	0.787	0.892	5

The performance summary (MCR = Misclassification rate, AUC = Area under the curve, Sens = Sensitivity, Spec = Specificity, No. Components = the number of components used in the model) of each classifier for both the full dataset (“full”) and the trimmed dataset (“trimmed”) that underwent variable selection using a univariate moderated t-statistic. These are mean values based on 1000 bootstrap samples for each dataset except the OC data which used 200 bootstrap samples.