Skip to main content

Table 3 Results from decision tree analysis on Post

From: Methods for detecting methylation by SNP interaction in GAW20 simulation

 

Decision tree (importance)

Random forest (importance rank)

Predictor

Training

Validation

Mean (SD)

Median

rs736004

1.0000

0.9645

33.86 (27.75)

22

rs9661059

0.9989

1.0000

14.8 (13.29)

10

rs10828412

0.9843

0.9860

34.75 (24.16)

25.5

rs1012116

0.8998

0.8922

20.25 (15.08)

14

rs4399565

0.7838

0.6970

23.27 (15.25)

19

cg18772399

0.6793

0.5597

4.415 (1.72)

4

cg00000363

0.5758

0.4951

4.085 (1.75)

4

cg10480950

0.5348

0.3802

4.59 (1.91)

4

cg00045910

0.4441

0.3508

4.715 (1.74)

5

cg01242676

0.4333

0.2718

4.7 (1.79)

5

Pre

–

–

1 (0)

1

Age

–

–

5.075 (1.74)

5

Sex

–

–

87.625 (10.21)

90

  1. Decision analysis consisted of all 200 replications combined, and then separated into training (n = 54,334), validation (n = 40,769), and test (n = 40,897). Random forests were performed for each replicate separately, and each variable was ranked for importance. Pre, age, and sex were included as covariates