Table 6 Comparison of PIMA Indian results across different models.

From: Optimizing imbalanced learning with genetic algorithm

%

Metrics

SMOTE9

ADASYN11

GAN71

VAE63

SGA

EGA

SVMGA

20%

Accuracy

71.1

75.0

76.6

73.1

72.1

74.3

78.2

Precision

61.0

67.9

74.6

58.8

64.5

65.9

73.6

Recall

62.0

61.8

59.5

67.4

69.6

65.1

62.9

F1 Score

62.2

64.7

66.2

65.9

65.6

65.5

67.8

ROC AUC

69.1

71.8

73.4

72.0

71.8

72.0

74.4

40%

Accuracy

75.0

75.4

67.0

75.5

73.0

75.6

75.4

Precision

65.6

67.4

54.9

68.9

65.3

67.0

67.0

Recall

73.0

67.4

89.9

67.4

69.6

70.7

68.5

F1 Score

69.1

67.4

65.5

68.1

67.4

68.9

67.8

ROC AUC

74.5

73.4

69.7

74.2

73.2

74.5

73.7

60%

Accuracy

73.2

74.0

71.4

72.7

76.9

75.1

78.9

Precision

64.1

65.2

59.3

65.1

73.5

67.0

73.7

Recall

66.3

67.4

82.0

62.9

56.0

66.3

66.3

F1 Score

65.1

66.3

68.8

64.0

64.0

66.6

69.8

ROC AUC

71.5

72.4

73.4

70.9

71.8

72.9

75.7

80%

Accuracy

70.5

67.6

72.2

74.0

60.6

69.0

76.1

Precision

65.0

56.7

60.0

62.8

57.0

60.0

60.8

Recall

71.0

85.0

84.2

79.7

82.0

78.0

88.3

F1 Score

67.9

68.0

70.0

70.3

67.3

67.8

72.0

ROC AUC

70.0

72.0

74.5

75.1

67.0

72.0

75.7

100%

Accuracy

68.8

69.0

69.2

76.2

71.3

73.6

76.4

Precision

59.1

55.2

58.3

67.7

70.8

60.6

59.0

Recall

72.6

83.4

70.7

73.0

76.0

74.0

91.4

F1 Score

65.2

66.4

63.9

70.2

73.3

66.6

71.5

ROC AUC

71.0

71.4

69.5

75.6

72.0

74.0

76.3

  1. This table shows the performance metrics (accuracy, precision, recall, F1 score, and ROC AUC) for models trained with different synthetic data generation techniques (SMOTE, ADASYN, GAN, VAE, SGA, EGA, and SVMGA) on the PIMA Indian diabetes dataset at varying percentages of data usage.
  2. Significant values are in bold.