Yeah, my guess is that somewhere up to 5% of Verifiable observations are submitted as Unknowns and it’s only the valiant efforts of identifiers like you that keep the percent of Unknowns down so low when measured at any given moment.
No surprise that the ‘peaks’ are at the major taxonomic ranks. Interesting that >2/3 of the needsID pile is either at Genus or Species. Testament to the work of the ‘higher order improvers’ and also encouragement to those working at the finer end of the scale that it’s a good place for them to focus. I’m surprised that Family is only 8%, and Order only 4%.
If unknowns are coming in circa 5% then the fact that they are well below 1% of the pile is a great achievement.
I don’t understand. “Needs ID” depends on the ID rank (if above rank species, an observation will always need an ID, if not casual), on the “casual” status and on the “research grade” status. So the rationale is not simple, when plotting “Needs ID” rank by rank, because “Needs ID” and rank are not independant by definition. But I don’t know what you are looking for, or evaluating.
All “unknowns” (2.5 % of all observations) should be “needs ID”. Update: only all those not “casual” are “needs ID”.
I also don’t understand the substraction. What is this substraction?
I think any substraction is risky, because we could miss a subtlety. A direct request for counting the results would be better. (Then we can sum the counts for consistency checking).
I think that :
For Species, Hybrid, Subspecies, Variety, Form, Infrahybrid, I would consider, the percents of observations (at the rank considered) that are “casual”, that “need ID” and that are “research grade”.
For all other ranks, and for “unknown” observations, only 2 of the 3 categories: “casual” and “needs ID”.
I think it’s normal, if observation lacks some data user is more likely to not id too, new users like uploading without any id, then many iders mark cultivated plants without iding, it’s not optimal, but when one group of school kids upload a thousand of planted petunias, it’s fine to not id every duplicate and such.
Needs ID is everything that comes up by default in the identify view. It’s what most people base their identifying effort on. It includes everything that’s not Research Grade or Casual so it doesn’t depend on rank very much, any rank can be casual, and any rank below Family can become Research Grade. It’s interesting to see unknowns as a proportion of all observations, but I think it’s also helpful to see the proportion of unknowns as a fraction of ‘what we have left to do’.
I don’t know if it is still useful, but I make it available.
I can regenerate it later, or with an additional filter (for instance, only >5 year old observations), without additional effort (generated by a software).
There are many “unknowns” observations that are casual just because they lack observation date (not “verifiable”), although they have a location (which is much more important than a date?).
It’s a pity to miss identifying so many observations just because of that.
Is observation date that important?
Anyway, they have a submission date, which is better than no date at all.
Shouldn’t observations with missing observation date be considered “needs ID” (“verifiable”)?
Date is vital for phenology. When does it bloom, this year, last year, during the drought years, after the Polar Vortex …
Also problems from people battling internet connection, or loadshedding (ask me how I know - but mostly from comments, since my own photos are camera and uploaded thoughtfully much later)
We need to split Captive / Cultivated, better named honestly as Not Wild
from Casual again better named Lacking Data.
And Needs ID should be running independently from Wild / Not / Lacking. Been fighting for that since I landed on iNat.
Fortenately, I can push the “Casual” and the “Needs ID” to the same projects,
and let you append this filter to the URL, for reviewing only the “Needs ID”:
&quality_grade=needs_id
For instance, from this link (for identifying all the 61 x 30 observations in 3 projects):
you would obtain this other URL (only 15 x 30 observations):