Table 2 Predictive performances of the seven candidate learning models using microscopy-derived data.

From: Machine learning-based approaches for identifying human blood cells harboring CRISPR-mediated fetal chromatin domain ablations

Architecture

m00

m01

m10

m11

FCD-HT precision

FCD-HT recall

FCD-HT f-value

KU-812 precision

KU-812 recall

KU-812 f-value

Accuracy

AUC

Type 1 Depth 2 (T1D2)

79

82

68

104

0.559139785

0.604651163

0.581005587

0.537414966

0.49068323

0.512987013

0.54954955

0.553933

Type 1 Depth 4 (T1D4)

97

64

45

127

0.664921466

0.738372093

0.699724518

0.683098592

0.602484472

0.640264026

0.672672673

0.697783

Type 1 Depth 6 (T1D6)

76

85

94

78

0.478527607

0.453488372

0.465671642

0.447058824

0.472049689

0.459214502

0.462462462

0.500469

Type 2 Depth 2 (T2D2)

94

67

55

117

0.635869565

0.680232558

0.657303371

0.630872483

0.583850932

0.606451613

0.633633634

0.529684

Type 2 Depth 3 (T2D3)

103

58

53

119

0.672316384

0.691860465

0.681948424

0.66025641

0.639751553

0.649842271

0.666666667

0.642514

Type 2 Depth 4 (T2D4)

106

55

35

137

0.713541667

0.796511628

0.752747253

0.75177305

0.658385093

0.701986755

0.72972973

0.73653

Type 2 Depth 5 (T2D5)

136

25

42

130

0.838709677

0.755813953

0.795107034

0.764044944

0.844720497

0.802359882

0.798798799

0.874621

Back to article page