Skip to main content
. 2010 Mar 11;26(9):1192–1198. doi: 10.1093/bioinformatics/btq107

Table 2.

Mean error rate percentage using 10-CV (10 validation trials) obtained on both datasets X*Z together

Gene selection t-test (1) RF (2) sPLS (3)
Prostate PLS-RF 28.98 (4.16)
cforest 24.65 (4.91)
multinom 27.09 (2.47) 26.33 (3.03) 26.96 (4.74)
ME-indep 29.24 (4.44) 28.86 (5.53) 31.64 (3.48)
ME-multinom 26.32 (4.77) 25.82 (4.74) 27.59 (5.62)
ME-loc 25.44 (3.70) 26.83 (2.44) 23.54 (2.61)
Breast PLS-RF 28.20 (1.86)
cforest 31.15 (1.60)
multinom 27.89 (1.69) 31.33 (1.53) 28.59 (2.02)
ME-indep 27.81 (1.47) 30.19 (1.73) 27.77 (1.28)
ME-multinom 27.65 (1.45) 29.88 (1.80) 27.89 (1.46)
ME-loc 27.85 (1.51) 29.92 (1.25) 27.26 (1.16)
CNS PLS-RF 38.68 (2.18)
cforest 38.33 (6.90)
multinom 33.67 (5.49) 36.83 (7.51) 37.17 (4.16)
ME-indep 43.83 (6.5) 42.81 (6.57) 41.83 (4.81)
ME-multinom 33.00 (6.08) 39.67 (6.08) 35.67 (6.34)
ME-loc 31.67 (6.98) 35.16 (5.63) 35.33 (5.37)

Classification performances which were as good or better than X* or Z alone (see Table 1) are indicated in bold (p* = 5).