Skip to main content

Table 3 Comparison of prediction accuracy for three datasets

From: Timing shift-based bi-residual network model for the detection of electricity stealing

Dataset

ESC

TS-BiResNet (%)

Bi-ResNet (%)

LSTM (%)

GRU (%)

CNN-LSTM (%)

D1

0.6

98.50

92.35

91.67

90.33

90.67

 

0.7

98.15

91.50

84.67

87.00

87.00

 

0.8

90.85

81.50

84.33

85.33

85.67

 

0.9

81.96

65.92

75.00

54.67

72.33

D2

0.6

83.08

82.67

NONE

NONE

73.50

 

0.7

82.60

80.00

NONE

NONE

70.25

 

0.8

78.27

73.25

NONE

NONE

58.67

 

0.9

63.94

60.93

NONE

NONE

52.33

D3

0.6

95.85

93.67

91.00

92.67

87.33

 

0.7

93.15

90.25

87.67

85.67

81.00

 

0.8

90.65

81.27

68.93

75.33

66.33

 

0.9

67.54

60.77

62.67

56.33

52.67