Skip to main content

Table 1 Simulation results for the empirical Bayes method with known groupings (blocks)

From: Empirical Bayes method for reducing false discovery rates of correlation matrices with block diagonal structure

 

Pearson block

Corpcor block

EB block

EB

Mean n=10

5.67

20.41

4.08

9.23

Sd n=10

1.49

1.44

1.64

1.72

Mean n=15

4.10

17.99

2.81

7.59

Sd n=15

0.98

0.93

1.05

1.13

Mean n=20

4.08

18.42

2.69

7.39

Sd n=20

0.76

0.80

0.75

0.90

  1. We calculate the Frobenius norm between the estimated and true matrix for elements within the blocks. The estimates for the full matrices are the same for the Pearson and the Corpcor methods by construction, in contrast there is a small increase in the Frobenius norm for the EB method. This is due to the fact that for the Pearson and Corpcor method we set all elements outside the known blocks to zero whilst the EB method sets these elements to zero in the prior matrix, rather than the final estimated matrix