Skip to main content

Table 4 Estimated error rates (ATE) and standard deviations for an increasing number of features on tumor vs. metastases data (average over 50 experiments), for each of the ranking methods.

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

# genes

E-RFE

SQRT-RFE

10

17.9 ± 5.2

17.3 ± 6.0

20

16.7 ± 2.9

16.1 ± 3.9

50

16.4 ± 3.8

16.2 ± 3.5

100

16.2 ± 3.7

15.8 ± 3.5

300

13.7 ± 4.3

13.6 ± 4.4

500

13.4 ± 4.4

13.6 ± 4.8

1000

13.3 ± 5.2

13.3 ± 5.2

16063

All features: 12.9 ± 5.2