posterior
posterior copied to clipboard
pareto diagnostics return named values, preventing renaming
Currently the pareto_* summary functions return named vectors without the preceding pareto_. e.g. pareto_convergence_rate returns a numeric named convergence_rate.
This was done as the function names were deemed too long for column names. However, this also has the side effect that the columns can not be renamed in summarise_draws (and convergence_rate is still quite long). I think all other built-in diagnostics that return single values are unnamed, so this inconsistency is a bit unexpected.
The functions names and lengths are:
pareto_convergence_rate (23 characters, still 16 without pareto_)
pareto_khat_threshold (21 characters, 14 without pareto_)
pareto_min_ss (13 characters, 6 without pareto)
pareto_khat. (11 characters, 4 without pareto_)
In contrast, the longest name for other built-in summary functions is rhat_nested (11 characters, also a recent addition), and others are shorter than 10 characters (e.g. ess_tail).
I see two options to fix this:
- shorten the function names where possible and then change to return unnamed values
- keep returning named values, but shorten the names for
convergence_rateandkhat_threshold(maybeconvrateandkhat_thresh?)