[Bug:] Multiple entries age comp entries for single fleet/year combo with petrale

[kellijohnson-NOAA]

Describe the bug

@iantaylor-NOAA I am trying to update the case studies to use the newest version of dev and I found a bug that we were not catching before. For multiple years of age-composition data within the SS3 data file there is more than one row of age comps for fleet 1. I am wondering how we decide between them or do we combine them? Years with invalid data include the following: 1999, 2009, 2010, 2011, 2012, 2013, 2014, 2015, 2016, 2017, 2018, 2019, 2020, 2021, and 2022. One idea I had is to just pick the ageerr row with the higher sample size but I am not sure if that is what you want to do.

This was caught within purrr but lacked an informative error message. I am working on a new branch of NOAA-FIMS/FIMS where I add in a cli abort call to inform the user that there are errors in their dimensions.

Describe how to reproduce the bug

Inside the petrale case study you can run the following code to see the data that I am referring to

petrale_input$dat$agecomp |>
  dplyr::filter(fleet %in% c(1))

or

petrale_input$dat$agecomp |>
  dplyr::filter(fleet %in% c(1), year == 1999)

to see a specific instance of what I am talking about.

Screenshots

No response

Which OS are you seeing the problem on?

No response

Which browser are you seeing the problem on?

No response

Which version of FIMS are you seeing the problem on?

Current dev with last commit of e3b75b1

Additional context

No response

[Rick-Methot-NOAA]

This data file has multiple ageerror methods defined. These rows in question are not duplicates; they are for different ageerror methods.

[kellijohnson-NOAA]

It is not a bug in SS3 it is a bug in this FIMS case study because FIMS will not accommodate multiple rows for a single fleet/year combination. There are no ageing-error definitions in FIMS so we need to decide, specific to the petrale model, what to do with these multiple rows of data.

[Rick-Methot-NOAA]

I understood that to be the question

[iantaylor-NOAA]

@kellijohnson-NOAA, good question. The two solutions that seem best to me are (1) your suggestion to pick the row with the higher sample size, or (2) aggregate the rows into a single vector. The differences in ageing error matrices are not that large so the model results are probably pretty similar whichever of these options we use. I can help with this next week if you want.

[kellijohnson-NOAA]

I implemented (1) in https://github.com/NOAA-FIMS/case-studies/compare/dev-wtatage-from-SS3 using some dplyr. I still need to look at the model comparisons though.