compare changes in likelihoods by fleet?

[iantaylor-NOAA]

The r4ss::SStableComparisons() function is useful for producing the tables showing the changes in likelihood by data type that we've shown so far, but it would often be useful to look at the finer scale changes in likelihood by fleet (the table equivalent of Piner Plot vs the main profile plot showing the sum across all fleets).

The list returned by r4ss::SSsummarize() function (as used in https://github.com/iantaylor-NOAA/Lingcod_2021/blob/main/models/model_STAR_requests.R#L31-L37 with model files on Google Drive) includes a data.frame likelihoods_by_fleet but the format is terrible (it's just rbinding the likelihoods_by_fleet table from SS_output):

     model           Label          ALL X1_Comm_Trawl  X2_Comm_Fix    X3_Rec_WA
194      1    Catch_lambda           NA   1.00000e+00  1.00000e+00  1.00000e+00
195      1      Catch_like  6.99642e-10   5.80240e-10  4.13638e-11  2.12453e-11
196      1 Init_equ_lambda           NA   1.00000e+00  1.00000e+00  1.00000e+00
197      1   Init_equ_like  0.00000e+00   0.00000e+00  0.00000e+00  0.00000e+00
198      1     Surv_lambda           NA   1.00000e+00  1.00000e+00  1.00000e+00
199      1       Surv_like -1.62875e+02  -1.61206e+01 -1.48465e+01 -5.23057e+01
200      1      Surv_N_use           NA   1.70000e+01  1.70000e+01  4.00000e+01
201      1     Surv_N_skip           NA   0.00000e+00  0.00000e+00  0.00000e+00

@kellijohnson-NOAA, if you have time at some point, could you work some tidyverse magic to provide a table of changes in likelihood for each fleet for the Survey (really index), Length, and Age data. For these models I think we can ignore all the rows with labels "lambda", "N_use", "N_skip").

This would help us understand how something like removing fixed-gear ages improves the fit to the fixed-gear length as opposed to all length data.

[kellijohnson-NOAA]

I can look at this now

[kellijohnson-NOAA]

@iantaylor-NOAA do we need this for the document or can we label it as an r4ss post mortem issue?

[iantaylor-NOAA]

Labeled "postmortem: r4ss" and added to project