I wrote a very rough function today to do the sampling from these data. Input is the wtatage.ss_new file, and it takes the fleets, years, sample sizes and CVs specified by the user to sample lognormally from the given data and write the results to a wtatage file that can be read into the EM.
I tested this with the cod model and surprisingly it ran without any modifications besides changing the maturity option to 5 (there is already a function change_maturity which I used in the package to do this automatically). SS did warn me about using length-based selectivity, but ran and seemingly converged.
At this point I think I need feedback from the empirical group to what the next steps are. As I see it, getting the hake OM/EM tested will be important, but now that should be easier since this function is now in the package under feature/data. After that is working fine, then the group can turn to how to improve and adapt the sample_wtatage function so that (1) its statistical sampling assumptions are more realistic, and (2) it is capable of doing the scenarios you're interested in testing.
Peter: I'm happy to show you how to use the function. From there you could work with someone (maybe Kot if Allan is still unavailable?) to get those models working. In the meantime, you could probably just use the flatfish model and start running your basic scenarios.
I wrote a very rough function today to do the sampling from these data. Input is the wtatage.ss_new file, and it takes the fleets, years, sample sizes and CVs specified by the user to sample lognormally from the given data and write the results to a wtatage file that can be read into the EM.
I tested this with the cod model and surprisingly it ran without any modifications besides changing the maturity option to 5 (there is already a function
change_maturitywhich I used in the package to do this automatically). SS did warn me about using length-based selectivity, but ran and seemingly converged.At this point I think I need feedback from the empirical group to what the next steps are. As I see it, getting the hake OM/EM tested will be important, but now that should be easier since this function is now in the package under feature/data. After that is working fine, then the group can turn to how to improve and adapt the
sample_wtatagefunction so that (1) its statistical sampling assumptions are more realistic, and (2) it is capable of doing the scenarios you're interested in testing.Peter: I'm happy to show you how to use the function. From there you could work with someone (maybe Kot if Allan is still unavailable?) to get those models working. In the meantime, you could probably just use the flatfish model and start running your basic scenarios.