Table 8 

The comparison of SVM vs. WV on DataG 

Level^{a} 
ReduceTest^{d} 
P(testdiff)^{e} 
ImproveRec^{f} 
P(recdiff)^{g} 


800 
36.36% 
1.16E17 
1.02% 
2.13E06 
600 
38.95% 
6.74E17 
9.49% 
2.14E62 
500 
39.51% 
8.72E21 
14.82% 
3.77E71 
400 
44.84% 
3.86E23 
20.83% 
1.68E79 
300 
49.75% 
6.86E25 
28.72% 
3.48E86 
200 
54.22% 
2.02E27 
36.75% 
3.70E91 
150 
54.83% 
9.37E30 
36.14% 
2.65E86 
100 
43.56% 
6.63E25 
33.61% 
4.42E75 
90 
42.35% 
1.85E26 
31.09% 
4.23E73 
80 
37.37% 
7.35E25 
29.08% 
3.79E67 
70 
32.23% 
1.20E20 
26.54% 
9.22E63 
60 
27.79% 
1.16E20 
24.39% 
1.24E61 
50 
23.47% 
8.64E15 
21.80% 
1.83E53 


^{a }Level: The number of features selected in each recursive step. ^{d}ReduceTest: Relative reduction in the mean test error rates of SVM comparing to that of WV, calculated as: (average TestError_{WV } average TestError_{RSVM})/(average TestError_{WV}). ^{e }P(testdiff): The pvalue of the observed differences in test error rates, by paired ttest. ^{f }ImproveRec: Relative improvement in the proportion of recovered informative genes by RSVM comparing to that by WV, calculated as: (average #REC_{RSVM } average #REC_{WV})/(average #REC_{WV}), where #REC represents the number of recovered true informative genes with the method stated in the subscript. ^{g }P(recdiff): The pvalue of the observed difference in proportion of recovered informative genes, by paired ttest. 

Zhang et al. BMC Bioinformatics 2006 7:197 doi:10.1186/147121057197 