samaperrin
commented
11 months ago Hi Jens, this is something I'm actually fixing at the moment. We're introducing specialised processing scripts for datasets, including the national insect monitoring, which take into account individual characteristics. CHanges will be pushed withint he next week, though they'll be on the forked version under my repository (we've submitted this version for the Ebbe Nielsen Challenge so it needs to stay untouched until October).
Thanks for taking an interest! Sam
Hi, I stumbled upon this repo and the shiny-app. Very nice!
Just a note regarding the data from the national insect monitoring in norway. In the occurrence map, null-records for this dataset are shown even when you deselect "show absences". Perhaps just a plotting error, or indicative of an interpretation error in the data that goes into the models? According to my records, I can't see that we have observed any of the long-horn beetles in Trøndelag, still it looks like all our sites are plotted in the occurrence map.
In this dataset, we don't measure or report "individualCount". Instead we report an "organismQuantity" of "organismQuantityType": 'DNA-reads', i.e. we measure how many copies of the DNA-strands we have in each sample. Looking through the code, it seems that individualCount is inferred to be 1 when this column is missing, at least in the shiny-app. Perhaps this is the source of the plotting error?
Lastly, for this dataset when inferring absence-observations, note that we have so far have registered over 20 000 "species", where many of them is very rarely found (but only the ones that can be attributed to a species name with reasonable confidence is reported on GBIF). Occurrence and detection probabilities in a single location is probably pretty low for many species. Not necessarily a problem for your models, but I'd thought I'd mention it.
Again, interesting work and app! /Jens Åström