Skip to main content
. Author manuscript; available in PMC: 2010 Feb 25.
Published in final edited form as: Genet Epidemiol. 2008 May;32(4):350–360. doi: 10.1002/gepi.20309

TABLE II.

Summary of predictability of Random Forests and RuleFit with five-fold cross-validation using 17 risk factors and 287 tagSNPs

Random Forests
RuleFit
Data set 1
Data set 2
Data set 1
Data set 2
AUC Sensitivity
(%)
Specificity
(%)
AUC Sensitivity
(%)
Specificity
(%)
AUC Sensitivity
(%)
Specificity
(%)
AUC Sensitivity
(%)
Specificity
(%)
Subset 1 0.746 80.8 60.0 0.843 89.1 61.5 0.690 71.2 60.0 0.746 73.9 61.5
Subset 2 0.746 70.7 71.0 0.675 71.2 60.0 0.681 73.2 61.3 0.614 71.2 50.0
Subset 3 0.778 79.6 61.1 0.718 74.5 61.9 0.667 70.4 61.1 0.696 76.5 61.9
Subset 4 0.752 75.0 61.5 0.718 71.9 69.2 0.692 68.8 65.4 0.739 78.1 61.5
Subset 5 0.777 77.5 71.4 0.766 81.3 68.2 0.786 85.0 71.4 0.665 68.8 59.1
Mean 0.760 76.7 65.0 0.744 77.6 64.2 0.703 73.7 63.8 0.692 73.7 58.8
STD 0.016 4.0 5.7 0.064 7.6 4.2 0.047 6.5 4.7 0.054 3.8 5.1

SNP, single nucleotide polymorphism; AUC, area under the curve.