Skip to main content

Table 4 Ranking of all combinations of methods

From: Comparative evaluation of set-level techniques in predictive classification of gene expression samples

Rank

Methods

  

Avg Subrank

 

Sets

Rank. algo

Aggrgt

 

1

1:10

Global

None

15.3

2

1:10

Global

SetSig

15.7

3

1

Global

None

16.3

4

1:10

GSEA

None

16.7

5

baseline (all genes used)

16.8

6

1:10

Global

SVD

17.0

7

1:10

SAM-GS

None

17.2

8

1:10

SAM-GS

SetSig

17.6

9

1:10

Global

AVG

18.6

10

1

Global

SVD

19.4

11

1:10

GSEA

SetSig

19.9

12

1:10

GSEA

SVD

20.1

13

1:10

SAM-GS

SVD

20.8

14

1:10

GSEA

AVG

22.1

15

1

Global

SetSig

22.2

16

1

SAM-GS

None

23.0

17

1

SAM-GS

SetSig

23.8

18

1

GSEA

None

23.9

19

1

Global

AVG

24.6

20

1:10

SAM-GS

AVG

25.5

21

1

GSEA

SVD

26.7

22

1

GSEA

SetSig

26.8

23

1

SAM-GS

SVD

28.3

24

1

SAM-GS

AVG

30.3

25

1

GSEA

AVG

30.9

  1. Ranking of all combinations of methods in terms of average subrank. Subranking is done on each of the 150 combinations of 30 datasets and 5 learning algorithms by cross-validated predictive accuracy. Column descriptions are as in Table 3.