Table 2 Performance of 18 regression modeling methods on the three HMX exposure datasets assessed by coefficient of determination (R², mean ± standard deviation, n = 10) estimated from ten runs of 10-fold cross-validation with values of the best performing method shown in bold

Predictor size (gene #)

6

6

10

 Multivariate

0.53 ± 0.15

0.52 ± 0.15

0.58 ± 0.15

 Robust

0.66 ± 0.12

0.72 ± 0.09

0.79 ± 0.02

 Ridge

0.67 ± 0.10

0.70 ± 0.11

0.81 ± 0.02

 LASSO

0.69 ± 0.10

0.72 ± 0.10

0.81 ± 0.04

 Elastic net

0.72 ± 0.09

0.71 ± 0.11

0.82 ± 0.03

 SVR

0.70 ± 0.10

0.65 ± 0.09

0.81 ± 0.05

 Stepwise

0.67 ± 0.07

0.66 ± 0.11

0.79 ± 0.05

 Ridge Polynomial

0.63 ± 0.11

0.73 ± 0.08

0.76 ± 0.05

 Ridge Exponential

0.68 ± 0.08

0.68 ± 0.09

0.79 ± 0.04

 Ridge Gaussian

0.51 ± 0.16

0.56 ± 0.14

0.66 ± 0.06

 SVR Polynomial

0.69 ± 0.11

0.64 ± 0.11

0.79 ± 0.06

 SVR Gaussian

0.65 ± 0.09

0.60 ± 0.10

0.73 ± 0.10

 SVR Sigmoid

0.48 ± 0.15

0.49 ± 0.15

0.68 ± 0.12

 Nadaraya-Watson

0.68 ± 0.09

0.67 ± 0.09

0.80 ± 0.04

 Inverse

NA

NA

NA

 Loglog

NA

NA

NA

 Regression Tree

0.56 ± 0.15

0.61 ± 0.14

0.65 ± 0.13

 Random Forest

0.55 ± 0.16

0.60 ± 0.13

0.69 ± 0.10