Feature request: Comparing Bayesian R2

[matticervin]

It would be great if there was a feature that could compare Bayesian R2 for (1) variables within a network and (2) the same variable across samples.

[donaldRwilliams]

I will add this soon ! Turns out we have an in progress paper looking at just this for Bayesian R2.

You can compare with the following:

# data
Y <- ptsd[,1:5]

fit <- estimate(Y)

r2 <- predictability(fit)

# compare node 1 vs node 2
diff <- r2$scores[[1]] - r2$scores[[2]]

hist(diff)

quantile(diff, probs = c(0.05, 0.95))

[matticervin]

This worked! Thanks a lot!

For others that are new to this, I simply did what Donald suggested and will report:
mean(r2$scores[[1]])
mean(r2$scores[[2]]
mean(diff)
quantile(diff, probs = c(0.05, 0.95)) for credibility interval
mean(diff > 0) for posterior probability