g3_fit does not report g3l_catchdistribution(obs_data in weight

[vbartolino]

When a data component fitted with g3l_catchdistribution has obs_data in weight rather than number, it seems that it does not get reported as part of the g3_fit. Any suggestion?

[lentinj]

Looks like it only looks for numeric reports:

gadgetutils/R/g3_fit.R

Lines 127 to 132 in fb3a203

	if (any(grepl('^cdist_.+__num$', names(tmp)))){
	## To-do: add in age and length attributes from stock objects
	## Merge together catch distribution observations and predictions
	dat <-
	tmp[grep('^cdist_.+__num$', names(tmp))] %>%

Changing the regexes from '^cdist_.+__num$' to '^cdist_.+__(num|wgt)$' should make it spring into life, but then you won't know the units of the output of g3_fit().

However, looks like surveyindices doesn't report the unit either so maybe that's okay? Not sure.

[vbartolino]

This seems to do the job (at least the data component is returned)

  if (any(grepl('^cdist_.+__num$|^cdist_.+__wgt$', names(tmp)))){
  
    ## To-do: add in age and length attributes from stock objects
    ## Merge together catch distribution observations and predictions
    dat <- 
      tmp[grep('^cdist_.+__num$|^cdist_.+__wgt$', names(tmp))] %>%
      purrr::map(as.data.frame.table, stringsAsFactors = FALSE) %>%
      dplyr::bind_rows(.id = 'comp') %>%
      dplyr::mutate(data_function = gsub('(cdist)_([A-Za-z]+)_(.+)_(model|obs)__(num|wgt), '\\2', .data$comp),
                    #type = gsub('(cdist)_([A-Za-z]+)_([A-Za-z]+)_(.+)_(model|obs)__(num|wgt), '\\3', .data$comp),
                    #fleetnames = gsub('(cdist)_([A-Za-z]+)_([A-Za-z]+)_(.+)_(model|obs)__(num|wgt)', '\\4', .data$comp),
                    origin = gsub('(cdist)_([A-Za-z]+)_(.+)_(model|obs)__(num|wgt)', '\\4', .data$comp),
                    name = gsub('(cdist)_([A-Za-z]+)_(.+)_(model|obs)__(num|wgt)', '\\3', .data$comp),

[willbutler42]

This is fixed here: 798d644. As Jamie noted, we do not know the units at the moment, but this can be added if you think it would be useful @vbartolino ?

[vbartolino]

Knowing the unit seems secondary to me (at least as I think now):

• in some cases (ie prey-species in the stomachs) I'm using proportions
• the importance is that observed and predicted are computed correctly from weights or numbers (as it is now, thanks for quick fix)
• in general the user knows the data unit