Skip to main content

Table 6 Estimated error rates, standard deviations, and optimal number of features (ATE: mean of test errors, from VAL procedures, averaged on B = 50 experiments; : average optimal number of features).

From: Entropy-based gene ranking without selection bias for the predictive classification of microarray data

 

Colon

Lymphoma

Tumor vs. met.

Method

ATE (%)

ATE (%)

ATE (%)

E-RFE

17.2 ± 7.6

70 ± 24

3.8 ± 4.3

80 ± 9

16.6 ± 2.9

70 ± 28

RFE

18.1 ± 8.4

70 ± 27

3.9 ± 4.1

80 ± 7

-

-

SQRT-RFE

18.4 ± 8.4

70 ± 27

4.3 ± 4.6

80 ± 9

16.2 ± 4.2

70 ± 25

CC

22.0 ± 10.1

50 ± 31

6.0 ± 4.8

80 ± 14

-

-

TT

20.9 ± 9.4

45 ± 30

5.7 ± 4.5

80 ± 11

-

-