Skip to main content

Table 1 Mean and standard deviation (SD) values of the balanced accuracy and Matthews Correlation Coefficient (MCC) for all five learning algorithms in 20 bootstrap test samples. The best values in each fold category are underlined with the overall best in bold

From: Prediction of novel mouse TLR9 agonists using a random forest approach

Algorithm

Cross-validation

Mean balanced accuracy

SD balanced accuracy

Mean MCC

SD MCC

RF

5-fold

77.8%

0.08

0.57

0.15

GBM

5-fold

76.8%

0.07

0.55

0.12

SDA

5-fold

74.6%

0.08

0.50

0.14

SVM

5-fold

77.1%

0.08

0.55

0.16

NN

5-fold

74.1%

0.07

0.50

0.13

RF

10-fold

78.9%

0.06

0.60

0.11

GBM

10-fold

77.7%

0.05

0.57

0.10

SDA

10-fold

75.8%

0.06

0.53

0.11

SVM

10-fold

78.4%

0.05

0.58

0.11

NN

10-fold

72.9%

0.05

0.48

0.10

RF

15-fold

77.0%

0.06

0.55

0.11

GBM

15-fold

76.9%

0.06

0.55

0.11

SDA

15-fold

73.5%

0.06

0.49

0.11

SVM

15-fold

76.3%

0.05

0.53

0.11

NN

15-fold

72.6%

0.07

0.47

0.15

RF

20-fold

79.7%

0.05

0.61

0.09

GBM

20-fold

78.5%

0.07

0.58

0.12

SDA

20-fold

76.1%

0.08

0.54

0.14

SVM

20-fold

75.4%

0.05

0.52

0.09

NN

20-fold

74.9%

0.07

0.52

0.13