Skip to main content

Table 2 Multivariate outliers simulated study

From: Enabling network inference methods to handle missing data and outliers

Outliers

Method

BM1

BM2

BM3

BM4

BM5

  

P

R

P

R

P

R

P

R

P

R

1 %

MIDER

66.7 %

66.7 %

84.9 %

88.2 %

86.9 %

82.0 %

88.3 %

88.8 %

99.9 %

99.9 %

 

MIDER+OS

100 % ∗

100 % ∗

91.6 % ∗

94.6 % ∗

90.8 % ∗

88.6 % ∗

91.5 % ∗

91.2 % ∗

98.6 %

98.6 %

5 %

MIDER

65.7 %

65.7 %

83.7 %

86.8 %

83.0 %

74.7 %

79.6 %

83.1 %

98.6 %

98.6 %

 

MIDER+OS

91.0 % ∗

91.0 % ∗

86.9 % ∗

90.5 % ∗

85.6 % ∗

80.2 % ∗

82.8 % ∗

85.1 % ∗

97.3 %

97.3 %

10 %

MIDER

64.0 %

64.0 %

80.6 %

83.2 %

70.1 %

69.7 %

70.4 %

77.3 %

98.8 %

98.8 %

 

MIDER+OS

73.0 % ∗

73.0 % ∗

82.6 % ∗

85.8 % ∗

72.8 % ∗

71.1 % ∗

73.0 % ∗

79.2 % ∗

98.2 %

98.2 %

  1. Precision and recall of the network inference of MIDER and MIDER + the outlier scheme (MIDER+OS)
  2. The asterisks mark the statistically better results of MIDER+OS (p-value <0.05)