Problems with unknowns

[slucey]

Unknown areas are not allocating the way they have in the past. This was first noticed during the SOE leads meeting when looking at the Bennet Index.

[slucey]

First issue was that revenue was not proportioned correctly. Unknown matches would proportion the landings but not the revenue. So an unknown matched to three known records would triple the revenue from that catch.

This issue has been resolved.

[slucey]

Second big issue is that unknown records would match with unknown variables. So say a record had three known variables and one unknown variable, it could match with known records with the same missing variable. This allows for easier matches that may not be correct. This was noticed with Eastern Oyster (NESPP3 789) showing up in the Gulf of Maine as more important than lobsters.

[slucey]

Coding it so it matches by complete cases before winnowing down variables. Still need to add a combination function to get different combinations of known variables. gtools::combinations is a promising tool to get this to work.