Feature Request: glance on a list of models?

[billdenney]

I often want to compare many models-- usually collected in a list-- to each other. I've currently implemented this in my PKNCA package with the non-exported AIC.list() function, but as I think of it more, broom (and broom.mixed) seem like a better home for the type of functionality I'm thinking of.

While I can accomplish the same thing with a call like the following (typed directly into GitHub; it may have bugs, but it hopefully indicates the intent):

model_list %>%
  lapply(FUN=glance) %>%
  bind_rows()

For me, the multiple-model comparison need is common and I'd like to have an interface like:

glance_list <- function(x, ...) {
  ret_list <- lapply(X=x, FUN=glance, ...)
  bind_rows(ret_list)
}

To make this work with models that don't converge, I'd also ideally like to have glance() methods for logical NA and NULL:

glance.logical <- function(x, ...) {
  if (length(x) == 1 && is.na(x)) {
    # Choose appropriate "model summaries" for NA, possibly an empty data.frame, perhaps:
    data.frame(convergence=FALSE)
  } else {
    stop("Cannot glance at a logical value other than scalar NA.")
  }
}

glance.NULL <- function(x, ...) {
  # Choose appropriate "model summaries" for NULL, matching NA
  glance.logical(NA, ...)
}

This relates to: billdenney/pknca#71 and billdenney/pknca#90

[alexpghayes]

I recommend using purrr::map_df(model_list, glance) for this. Does that work for your use case?

[billdenney]

@alexpghayes, That works well enough.

[alexpghayes]

Just realized that glance() currently returns an empty tibble for NULL values, but there isn't a glance() method for NA values, for models that aren't converging. Are there any model types / packages you work with that return NA for models that failed to converge?

[billdenney]

Thanks for the quick reply!

My own code does within the PKNCA package, and the empty tibble return is an issue there. I started the discussion of NA in #565. (It seemed out of place to continue that conversation here.)

[github-actions]

This issue has been automatically locked. If you believe you have found a related problem, please file a new issue (with a reprex: https://reprex.tidyverse.org) and link to this issue.