Create native Dirichlet multinomial family #1729
Conversation
|
Thank you! This PR looks good overall, I think. Why are not confident about posterior_epred? You should be able to test it by comparing its output against the known(?) mean of the dirichlet_mulinomial distribution. I currently don't have time to check this detail myself. |
|
Thanks @paul-buerkner for taking a look at the PR - much appreciated. I've now checked I also had a quick scan of the checks above in case the failed checks were related to the dirichlet_multinomial family - I don't think they are. |
| # @param eta the linear predictor (of length or ncol ncat) | ||
| # @param phi the dispersion parameter | ||
| # @param log return values on the log scale? | ||
| ddirichletmultinomial <- function(x, eta, phi, log = FALSE) { |
There was a problem hiding this comment.
does this accept the same kinds of inputs as brms::dmultinomial (just with an additional phi?). E.g. is the right dimensionality for x and eta taken care of?
| eta <- get_Mu(prep, i = i) | ||
| eta <- insert_refcat(eta, refcat = prep$refcat) | ||
| phi <- get_dpar(prep, "phi", i = i) | ||
| cats <- seq_len(prep$data$ncat) |
There was a problem hiding this comment.
the variables cats and alpha seem redundant here. Leftovers from previous versions?
| eta <- insert_refcat(slice_col(eta, i), refcat = prep$refcat) | ||
| dcategorical(cats, eta = eta) * trials[i] | ||
| } | ||
| # dirichlet part included in mu |
There was a problem hiding this comment.
I don't understand this comment
| @@ -587,6 +587,21 @@ posterior_epred_multinomial <- function(prep) { | |||
| out | |||
| } | |||
|
|
|||
| posterior_epred_dirichlet_multinomial <- function(prep) { | |||
There was a problem hiding this comment.
Since this is now the same as the one for multinomial (right?), perhaps it makes sense to just call it directly with a comment saying why this is valid?
Following discussion in #1728, I've had a go at creating a native Dirichlet multinomial family.
I've attempted to update the documentation of
brmsfamily, and added some tests too.I'm not confident that the
posterior_epred.Rfunction is correct, and probably should not have included it.Hopefully this is helpful - please let me know if you need anything more from me. And thanks again @paul-buerkner for all your help with this!