search

BenWilliams-NOAA / swo-tech

0 stars 1 forks source link

equation question #11

Open BenWilliams-NOAA opened 2 years ago

BenWilliams-NOAA commented 2 years ago

https://github.com/BenWilliams-NOAA/swo-tech/blob/d2359bac6090f637c2900fab94ce33771248d666/text/swo_length_tech_memo.rmd#L145

Jason asked: Where did Eq. 4.2 come from? I don't doubt that it is used sometimes, but I don't believe it is the best way to approach missing data and I am not sure we want to include this complexity in the memo.

It is referencing the lpop function https://github.com/BenWilliams-NOAA/swo/blob/master/R/lpop.R

pete-hulson commented 2 years ago

This was from Wayne's old script (originally written by Michael Martin). I totally agree that it's not the best way to deal with missing data, but will leave it to GAP to suggest an alternative. I'm OK with not including this level of detail in the journal version, but am slightly on the side that if we're providing the first documentation of these methods in the tech memo that it should be included. This would also give them something to reference if they ever decide to change it.

BenWilliams-NOAA commented 2 years ago

Added size comp sql file (though in .txt format) the no length hauls begins in lines 63-98, and then is called again in lines 230-239)

sizecomp_sql.txt

Jason-Conner-NOAA commented 2 years ago

Thanks Ben and Pete, @pete-hulson, I take your point about documenting this process and it would definitely be a step in the right direction for transparency. I am looking into whether this is also done in the Bering Sea SQL scripts.

I would be okay leaving this in if the topic is elaborated on a little more, and I will work on a sentence to add. My concern for the analysis, and for our estimates, is that filling "data" in this way does not add information to the estimate and possibly decreases the fidelity of that information. However, these are definitely corner cases and I don't expect there to be a noticeable difference in the analysis with or without.

pete-hulson commented 2 years ago

You bet - that'd be great to add a sentence in there, and without too much consternation I bet we could see how many cases this actually applies to, my guess is (as you mentioned) that this isn't used very frequently

BenWilliams-NOAA commented 2 years ago

It is rather limited instances (at least for GOA). As @Jason-Conner-NOAA said it is unlikely that there is a noticeable diference

Jason-Conner-NOAA commented 2 years ago

I confirmed with Lyle Britt that the design-based sizecomps for the Bering Sea do not include this extrapolation. Where catch numbers exist but no lengths are observed, the portion within that stratum is coded as length -9 in the SQL output files.

So the next question is how do we handle this bifurcation in the analysis and tech memo?

pete-hulson commented 2 years ago

Man, that's really what should happen in the GOA and AI... Would a quick sentence describing that this applies to the GOA and AI only and not the EBS suffice?