daauerbach
commented
1 year ago I appreciate this write up. I think the current param that can take a “vehicle/trailer” string could either be used in this way or at least provides an analog.
On Fri, Oct 7, 2022 at 3:45 PM Kale Bentley @.***> wrote:
The group needs to revisit how interview data are summarized/filtered based on the presence/absence of "vehicle count" and "trailer count" data.
At our last creel analysis work session (on Oct. 6th, 2022), we realized that any interview with a missing "vehicle count" or "trailer count" data (value = NA) had been converted to zero. Although likely not a major issue, this is incorrect. Our quick fix to this issue was to filter interviews to only include ones where "vehicle count" and "trailer count" were not equal to NA.
While this solved one problem, it created another. First, and most importantly, our data wrangling code will now not work for any data sets where "vehicle count" or "trailer count" were not collected. This includes all of the R5 Steelhead Tributary creel data. Second, this can result in the loss of other useful data. For instance, if the interview had catch and hours fished but not vehicle/trailer counts, the entire interview would be omitted. Again, the absolute number of interviews that are lost due to this filtering choice is likely low. However, we can find a better solution.
To fix this bug, we need to update the script to at a minimum deal with the first issue above. Related to issue #23 https://github.com/wdfw-fp/CreelPointEstimate/issues/23, the interview data wrangling code could be subject to some sort of user input value on what index effort count data were collected. For instance, if indirect counts were conducted where things (vehicles, trailers, boats) need to be converted to anglers via interview data (anglers per vehicles, trailers, boats) then we wrangle the data one way. However, if only direct counts were conducted during index effort counts, then we would just want to ignore any summarization based on the presence/absence of count data from interviews.
— Reply to this email directly, view it on GitHub https://github.com/wdfw-fp/CreelPointEstimate/issues/24, or unsubscribe https://github.com/notifications/unsubscribe-auth/AEAKV3MA2LACJYZB3UIQC7LWCCRX7ANCNFSM6AAAAAAQ76745M . You are receiving this because you are subscribed to this thread.Message ID: @.***>
The group needs to revisit how interview data are summarized/filtered based on the presence/absence of "vehicle count" and "trailer count" data.
At our last creel analysis work session (on Oct. 6th, 2022), we realized that any interview with a missing "vehicle count" or "trailer count" data (value = NA) had been converted to zero. Although likely not a major issue, this is incorrect. Our quick fix to this issue was to filter interviews to only include ones where "vehicle count" and "trailer count" were not equal to NA.
While this solved one problem, it created another. First, and most importantly, our data wrangling code will now not work for any data sets where "vehicle count" or "trailer count" were not collected. This includes all of the R5 Steelhead Tributary creel data. Second, this can result in the loss of other useful data. For instance, if the interview had catch and hours fished but not vehicle/trailer counts, the entire interview would be omitted. Again, the absolute number of interviews that are lost due to this filtering choice is likely low. However, we can find a better solution.
To fix this bug, we need to update the script to at a minimum deal with the first issue above. Related to issue #23, the interview data wrangling code could be subject to some sort of user input value on what index effort count data were collected. For instance, if indirect counts were conducted where things (vehicles, trailers, boats) need to be converted to anglers via interview data (anglers per vehicles, trailers, boats) then we wrangle the data one way. However, if only direct counts were conducted during index effort counts, then we would just want to ignore any summarization based on the presence/absence of count data from interviews.