Skip to main content

Table 3 Generalisation error rate of parsimony models over seven benchmark data using four information theoretic filters

From: iRDA: a new filter towards predictive, stable, and enriched candidate genes

 

mRMR

CMIM

FCBF

iRDA

 

%

#

%

#

%

#

%

#

Brain

6

4

8

5

4

4

4

3

CNS

18.33

2

18.33

2

16.67

6

13.33

3

Colon

6.45

5

9.68

2

9.68

2

8.06

2

Leukemia

0

4

1.39

5

1.39

3

0

3

Lung

1.1

2

0.55

5

0.55

4

0.55

3

Lymphoma

1.3

4

2.6

2

24.68

1

1.3

3

Prostate

3.92

5

4.9

6

4.9

5

3.92

3

avg

5.3

3.71

6.49

3.86

8.84

3.57

4.45

2.85

  1. %: misclassification rate; #: number of explored genes