Aki Vehtari
Aki Vehtari
@jpritikin thanks for pointing out the package, although it doesn't solve the problem which of the many tests to use.
Thanks for the proposal. The reasons for me why I haven't worked on vignette are 1. we should fix first the density estimate as it now is dropping near edges...
Hmmm... There's a problem that we wouldn't like to add more dependencies to other packages (even that I suggested couple packages), so we should carefully consider what to do with...
> @avehtari It would be great to have this. Is one of your students working on this? Yes. It takes some time to have it in vignette as it's not...
Can you try with larger n? Preferably the same n a used in bayesplot to generate the bunch of reference density lines
> @avehtari Let me know if this looks like what you're looking for. Yes! Now the average of yrep lines is constant, even if the variation is slightly bigger near...
I think you don't need that double loop, but you can just use directly those runif values for the density estimation for the reference lines.
> You were right, I removed one loop and it now takes ~10 secs for 600 obs Good > I can't think of anything else top of mind on how...
Thanks for the suggestion, this makes sense. This is feasible as it would not change the projection part, just the criterion for whether the model is good enough.
Checking if you would like to collaborate on this?