Expected delay variance using `tidybayes`

[athowes]

The tidybayes package via brms has functionality to get the expected prediction .epred.

What we might also like to do is to get the expected prediction variance. This is implemented in add_mean_sd but perhaps there is a more native way to do it.

[athowes]

From @seabbs:

• This is not in tidybayes
• Write a helper function yourself and do it from samples. See e.g. this question
• One option which is very generic (good) but wasteful (bad) is to:
  1. .epred distributional parameters
  2. Sample from the continuous distribution parameters some number of times
  3. Summarise samples by mean, median, SD etc.
  4. Use these

[seabbs]

I think in terms of a standard interface then the last (wasteful) workflow seems like a good thing to support as it will work for all families regardless of what has or hasn't been implemented. Once that is there we can circle back and think about 2/or pushing for it to get better ecosystem support.