Skip to main content

Table 9 The proposed cost-sensitive method with other methods in term of G.Mean

From: Improved cost-sensitive representation of data for solving the imbalanced big data classification problem

Dataset

g-mean (%)

S-MVML-LA

GMEB

Method proposed

Iris

85.49 (3)

96.05 (2)

97.30 (1)

Cancer

96.33 (1)

94.98 (3)

96.22 (2)

Ionosphere

85.79 (3)

90.50 (1)

87.88 (2)

Wine

90.62 (3)

97.16(2)

98.85 (1)

Pid

0 (3)

75.74(2)

76.41 (1)

Wdbc

99.26 (1)

95.92 (3)

97.51 (2)

Musk

85.71 (2)

81.06(3)

87.54 (1)

Dermatology-6

79.85 (3)

98.13 (2)

99.70 (1)

FuelCons

60.23 (3)

84.53 (2)

86.50 (1)

Movement_libras

53.70 (3)

85.06(2)

85.43 (1)

Sonar

78.90(3)

78.90 (2)

84.67 (1)

SPECTF

50(2.5)

50 (2.5)

75.93 (1)

Mnist

87.29 (3)

94.94 (2)

95.42 (1)

Colon

81.63(2)

67.88 (3)

82.55 (1)

Caltch101

70.12 (1)

66.09 (3)

68.26 (2)

DLBCL77

50 (3)

82.70 (2)

95.50(1)

Kddcup-rootkit-imap-vs-back

57.13(3)

99.97 (1)

99.97 (1)

Kddcup-buffer-overflow-vs-back

79.22 (3)

100 (1)

100 (1)

Kddcup-guess-passwd-vs-satan

81.15 (3)

96.73 (2)

100 (1)

Kddcup-land-vs-satan

77.74 (3)

100 (1)

100 (1)

Cleveland

0 (3)

32.97 (2)

40.51 (1)

Average rank

2.59

2.09

1.23

  1. The numbers in bold are the best accuracies achieved on the dataset