Update kriging apportionment method and functions #215
Conversation
…into test-branch
…into test-branch
- This pertains to the kriging apportionment operation
- Also rearranges how `compute_summed_aged_proportions` is called to calculate the aged proportions since the sex proportions can be calculated in paralell with the total aged proportions in this step.
|
Just a note: for this one is fine, but for the next ones please do it from your fork. |
| # ---- Drop unaged fish | ||
| # ==== !!! TODO: pending what FEAT says, weights associated with | ||
| # ==== missing ages should be added into the 'unaged' category. | ||
| # ==== This would further mean apportioning these into `weight_strata_aged_uanged` | ||
| specimen_data_filtered = specimen_data[ specimen_data.sex != 'unsexed' ].dropna( how = 'any' , subset = 'age' ) |
There was a problem hiding this comment.
This does 2 things: select only sexed fish and dropped unaged fish, right?
Are there other place where unsexed fish from station 1 is used? If not, maybe we could also drop unsexed fish from station 1 when specimen_df is first read in. Or is it that there are too many unsexed fish so it is not a good idea to just drop them at data loader?
There was a problem hiding this comment.
- Drop unaged fish in data loader
- Adding caveat doc page related to the treatment of unsexed and removal of unaged fish
- Which functions do the unsexed fish become relevant (e.g. length-weight regression)
- Print out in console in data loader: "x unaged samples removed from Station 2 dataset" or something to that effect
There was a problem hiding this comment.
Hey @brandynlucca : Thanks for the PR! I made various inline comments. Below are some overall comments:
Under the function apportion_kriged_biomass, the outputs weight_strata_aged_unaged and unaged_biomass appear unused after they are created. Since you went all the way to the end for the apportioning, this means that these variables can be removed, right?
For dropping unaged fish in station 1, how about dropping those fish when you read in specimen_df in the data loader? This way we don't have to worry about dropping those separately in each functions. It would be as if those fish samples never exist.
I guess the above may not be true for unsexed fish, but in compute_aged_weight_proportions the unsexed fish are also dropped (not selected, here). Should these be removed at the data loader too? Or because the number is larger and these samples are used in some other processing assuming the unsexed fish are assumed to follow the same male:female proportion, we should not drop them at data loader?
Lastly, there are two types of function names compute_* and calculate_* under biology.py -- is that intentional? Seems that compute_* are functions that are called by calculate_*?
This includes: