Skip to main content

Table 2 Comparison between iCluster-ensemble and competing methods

From: An ensemble of the iCluster method to analyze longitudinal lncRNA expression data for psoriasis patients

Method

Are feature selection and classifier construction separate

Size

Predictive error

iCluster-ensemble

Yes

20

20%

iCluster

Yes

20

46.47%

GEE-based screening

Yes

20

33.33%

EDGE

Yes

27

43.33%

LASSO

No

0.43a

46.67%

LASSO-ensemble

Yes

20

26.67%

  1. aSince LASSO builds up the final model simultaneously with feature selection, the sizes of final model differ in single LOO runs. Here, the average of the sizes over resulting 30 LASSO models is given. Predictive error corresponds to the leave-one-out error (LOO) rate