Research Article

Recursive Feature Selection with Significant Variables of Support Vectors

Table 7

The CV results on the breast cancer dataset.


Num. of genes
RSVM SVM-RFESVM-
CV2Mean SVCV2Mean SVCV2Mean SV

47.02 47.02 47.02
6060 45.94 47.02 46.88
5151 45.31 47.02 44.88
4378 44.55 47.02 42.84
3721 43.20 47.00 39.55
3163 41.98 47.00 37.41
2689 40.92 46.84 35.67
2286 39.63 46.78 35.18
1943 38.41 46.61 34.31
1652 37.47 46.06 32.43
1404 36.41 45.39 32.31
1193 35.31 44.80 30.90
1014 34.12 44.39 30.24
862 33.04 43.14 29.76
733 32.10 42.27 29.02
623 30.84 41.31 27.10
530 30.10 40.96 26.16
450 29.29 39.49 24.35
382 29.00 38.04 23.37
325 28.78 37.49 22.33
276 28.14 36.59 20.78
235 27.55 35.27 19.94
200 26.67 34.14 19.24
170 25.98 33.53 18.14
144 25.12 31.67 17.94
122 23.84 30.67 17.63
104 22.12 29.65 17.53
88 21.02 27.67 16.76
75 20.39 25.92 16.78
64 19.39 24.61 16.43
54 18.37 22.14 16.33
46 18.10 21.57 15.57
39 16.61 19.24 15.57
33 15.55 16.86 15.04
28 13.98 15.96 14.31
24 13.04 15.02 13.69
20 12.02 12.78 13.39
17 11.20 11.65 12.76
14 10.29 10.59 12.43
12 9.16 9.73 12.04
10 8.35 8.69 12.16
8 7.29 7.51 12.22
7 7.22 6.86 12.86
6 6.29 6.20 13.43
5 5.98 6.20 13.63

Num. of genes: the number of selected genes in the recursive steps. CV2: total cross-validation error rate (CV2 error rate). Mean SV: mean number of support vectors used in the cross-validation at each level.