Skip to main content

Table 5 Simulation summary of the estimated numbers of types

From: Learning Microbial Community Structures with Supervised and Unsupervised Non-negative Matrix Factorization

SNR

2

5

10

  

Mean

sd

Mean

sd

Mean

sd

+∞

3

2/3.4

0/1.0

5/3.1

0.3/0.3

8.9/3.5

1.9/1.3

4

 

2/3.8

0/1.5

5/3.3

0.3/1.2

8.9/3.7

2.2/2.0

2

 

2/3.7

0/1.4

5/3.1

0.3/0.3

8.9/3.7

2.2/2.0

1

 

2/3.6

0/1.4

5.1/3.0

0.5/0.2

8.5/3.8

1.9/2.0

+∞

6

2/6.4

0/0.8

5/6.4

0.5/0.7

9.8/6.6

1.2/1.2

4

 

2/6.4

0/0.6

5.2/6.6

0.8/0.9

9.5/6.4

1.3/0.6

2

 

2/6.4

0/0.6

5.2/6.6

0.5/0.8

9.4/6.3

1.5/0.5

1

 

2/6.3

0/0.6

5.0/6.6

0.5/0.8

9.4/6.2

1.1/0.5

+∞

9

2.4/8.4

1.5/1.5

5.1/8.9

1.7/0.8

8.7/9.2

1.8/0.8

4

 

2/8.2

0/1.7

5/9.8

0.4/1.1

8.8/9.2

1.8/0.6

2

 

2.4/9.1

1.5/0.7

7.0/9.3

1.5/0.8

11.0/10.9

1.2/1.4

1

 

2.5/9.0

1.1/1.1

6.9/8.7

1.8/0.9

9.7/10.8

1.7/1.8

  1. For example, the first entry 2/3.4 means when the true number of types is 2 for class 1, 3 for class 2 and SNR=+∞, the mean numbers of types our method chooses are 2 for class 1 and 3.4 for class 2