Table 6 Five-fold cross-validation and independent evaluation test results of the SVM method for general datasets with selected features.

From: PrESOgenesis: A two-layer multi-label predictor for identifying fertility-related proteins using support vector machine and pseudo amino acid composition approach

Datasets

λ*

Five-fold cross-validation test

Independent evaluation test

Accuracy (%)

Sensitivity (%)

Specificity (%)

MCC (%)

Accuracy (%)

Sensitivity (%)

Specificity (%)

MCC (%)

1

0.05

79.95

82.1

79.15

59.9

79.5

80.12

77.18

58.99

2

0.08

79.74

80.88

79.53

59.46

77.99

77.78

76

55.9

3

0.09

80.12

81.29

79.88

60.22

77.03

78.36

74.24

54.11

4

0.04

79.91

82.24

79.02

59.68

77.58

78.07

75.21

55.13

5

0.09

80.19

81.9

79.63

60.37

79.5

78.36

78.13

50.91

Average

0.07

79.98

81.68

79.44

59.92

78.32

78.53

76.15

55

  1. *The optimum λ parameter value of kernel function of SVM using a grid-search technique based on five-fold cross-validation. Also, the optimum parameter C value was obtained 100 in all of models.
Back to article page