| Pearson | Corpcor | EB | EB AIC |
---|
Mean n=10 | 19.27 | 23.48 | 17.65 | 17.86 |
Sd n=10 | 3.42 | 3.70 | 3.34 | 3.14 |
Mean n=15 | 12.30 | 17.45 | 12.72 | 13.89 |
Sd n=15 | 4.34 | 2.04 | 3.81 | 3.21 |
Mean n=20 | 10.52 | 17.52 | 11.46 | 13.19 |
Sd n=20 | 3.96 | 1.81 | 3.28 | 2.81 |
- We calculate the Frobenius norm between the estimated and true matrix for the full matrix. This shows a particular improvement when sample size is small (n=10) for the EB method, as the matrix is closer to the true matrix than for either the Pearson or Corpcor method. We also compare the result when the AIC method is used to estimate the threshold level (EB AIC). This method shows similar improvements when n=10 and comparable results to the other methods with the larger sample sizes. Indicating the AIC function has provided a reasonable estimate of the threshold level