Skip to main content

Table 5 Accuracy of prediction of glycosylation sites with random forest and naïve bayes algorithms

From: Prediction of glycosylation sites using random forests

 

Random Forest

Naïve Bayes

Dataset (size)

Correctly Classified Instances (%)

Sensitivity (%)

Specificity (%)

Matthews Correlation Coefficient

Correctly Classified Instances (%)

Sensitivity (%)

Specificity (%)

Matthews Correlation Coefficient

Ser

90.8

96.1

88.9

0.81

83.9

64.4

92.6

0.61

Ser + SA

91.1

95.5

89.6

0.82

82.3

60.5

92.3

0.58

Ser + Hydro

89.9

96.4

87.5

0.79

82.7

64.8

90.9

0.59

Ser + SS

91.7

96.3

90.1

0.83

82.4

62.9

91.3

0.58

Thr

92.0

93.6

92.4

0.84

86.8

74.8

93.3

0.70

Thr + SA

91.8

91.4

93.2

0.83

85.8

72.5

93.5

0.69

Thr + Hydro

91.1

91.8

92.2

0.82

85.9

73.0

93.3

0.69

Thr + SS

91.0

91.8

92.1

0.82

87.2

74.7

94.6

0.72

Asn

92.8

96.6

91.8

0.85

90.3

83.8

94.6

0.79

Asn + SA

94.0

95.7

94.3

0.88

89.3

81.9

94.5

0.77

Asn + Hydro

92.4

95.2

91.9

0.84

90.1

82.5

94.8

0.78

Asn + SS

93.2

96.4

92.4

0.86

89.3

79.8

94.9

0.76

  1. Hydro = Hydrophobicity data; SA = predicted surface accessibility; SS = predicted secondary structure.