α=kβ=G−k+1.
For k=2, G=3, our current best guess for the relationship between the determinant of the correlation matrix of the p-values and the MLEs for the Beta distribution is:
ˆα=ˆβ≈2−78√1−|P|.
Because the estimates for α and β are nearly identical, we conjecture that det(P) affects k and G equally, but this needs further exploration to confirm.
For each of the 100 response sets:
pathwayPCA::
R package to fit the logistic regression model of the random subtype against the first principal component of each pathway.For each of the 100 response sets:
pathwayPCA::
R package to fit the logistic regression model of the synthetic subtype against the first principal component of each pathway. (Recall that each platform has a different sample size.)Method | Median | Mean | Q3 |
---|---|---|---|
(1) | 0.055 | 0.062 | 0.080 |
(2) | 0.060 | 0.058 | 0.070 |
(3) | 0.050 | 0.051 | 0.060 |
(4) | 0.048 | 0.051 | 0.065 |
Stacked | 0.043 | 0.051 | 0.070 |
Comments on Flexibility