#IdentiFriday is the happiest day of the week

I :heart: Fridays
  • :raised_hand:

0 voters

There’s yet another school group active in Hong Kong, I’m okay with marking stuff on accounts with 70 observations, but 800 is just too much, so please, help marking all the duplicates and potted plants they upload. This region is very problematic in terms of uncontrolled groups, so maybe there should be a separate day in a week to deal with them alone.

2 Likes

i did a few, but there are a ton. if there’s time, i’ll come back and do more later.

2 Likes

I saw that, thank you!

I’ll write to the project owners. It’s probably too late to stop today’s current wave, but maybe it will help mitigate it in the next few weeks, which is how long at least one project is going for.

5 Likes

New Friday, new bingo card!

3 Likes

Is it possible to query a table with each row having a combination of iconic taxa and ID category? I’m already getting a lot of value out of being able to see that some users do, say, 100% plant IDs, while others do 50% birds, 25% insects, 20% mammals, 5% other stuff. And some users are 100% supporting IDs, and others 40% supporting, 60% leading + improving. But I would like to see whether some users are doing, for example, 100% supporting IDs in some taxa and some leading and improving in others.

Also, tables on which rows can be sorted by column would be nice.

P.S.: I noticed a user with 1 out of 6 identifications categorized “Other”: thebirdnerd

3 Likes

I don’t know about all time, but you can look up % of taxa in your year report, so you can find it out for every person. Or you want all people in a table, but separately?

1 Like

in the existing single-variable tables, the count for each category of each variable gets queried individually, requiring 14 requests (for 14 categories) in the iconic taxon table and 5 requests (for 5 categories) in the identification category table.

it’s possible to create a crosstab of the two variables, but it would require 70 requests (14 * 5 categories) to produce the results. so i hesitate to make something that would make it easy to get this data (since it’s kind of a clunky process). you could definitely create your own crosstab by iterating through and combining data from 5 variants (leading, improving, supporting, maverick, all) of the iconic taxon table though (or 14 variants of the identification category table).

if we had direct access to the database, then we might be able to get this data more efficiently, but i think regular users are stuck with the inefficient process.

1 Like

It’s exactly the automation of all that clunkiness which I desire. If it helps, I’m mostly interested in leading+improving vs supporting, so “only” 28 queries, and the automated calculation of percentages (and sorting of rows, if I can get it). Maybe group up Protozoa and Chromista with “Other Animals” as well to save another 4 queries. I’m not sure how much it’s worth to save that few queries to the API, given how many images iNaturalist serves. And I already reload my dashboard dozens of times per day, which sends off 100+ queries, many of them for images.

2 Likes

hmmm… sorry, making this kind of thing still doesn’t seem like a good ratio of pro vs con from my perspective. so i’m not going to do it. but maybe someone else will have no qualms about making such a thing? (the code for the existing 2 pages is located at https://github.com/jumear/stirfry, if it’s easier to adapt rather than build from scratch.)

in the meantime, you accomplish what you’re looking for just by loading 2 variants of the iconic taxon summary. for example:

https://jumear.github.io/stirfry/iNat_id_counts_by_iconic_taxa.html?own_observation=false&user_id=1722768&category=leading,improving

https://jumear.github.io/stirfry/iNat_id_counts_by_iconic_taxa.html?own_observation=false&user_id=1722768&category=supporting

2 Likes

I have a suspicion that the “person” I’m supporting is the computer vision!

3 Likes

Just curious, why do you want to know?

2 Likes

Many identifiers haven’t put much about themselves on their profile, so when I see that someone has specialized in one taxa or another or identified a wide variety of taxa it gives me a better idea of what they’re interested in, which sometimes means the difference between me tagging them or not.

Finding out that someone has done 98% supporting IDs even though they’re the top identifier for a taxon usually means I won’t tag them.

6 Likes

But you can learn it without table of their ids, I’d say the only way to learn about their competence is to look at their ids, you can add tons of wrong leading ids that nobody just critiqued yet, or only id when you’re tagged because you’re busy, so have only supporting ids, maybe the main thing “needed” is to find out what kind of mistakes they do and how often, which table can’t show.

Ah, I was wondering why there were hundreds of observations of trees coming in from Hong Kong marked as Ginkgo when they obviously weren’t.

4 Likes

Indeed, I can learn much more about an identifier by examining their identifications in detail. In practice, I only have time to view a subset of their IDs, and usually not a properly random sample. I have a reasonably good idea of the skill levels of other active identifiers who are identifying the same things I am (mostly spiders and hover flies), because I’ve viewed >100,000 observations, many of which have their IDs.

But for low-volume identifiers and people identifying mammals, for example, there is not enough time for me to view all their IDs or even a representative sample. Being able to view the complete, automatically computed statistics about an identifier’s activity in less than a minute is a very, very efficient use of my time. High ratio of information-learned to time-expended. It’s true that the aggregate statistics can occasionally be misleading, but it isn’t an efficient use of my time to seek out the perfect person to ask for help instead of someone with a high probability of being good enough. I shouldn’t spend more time trying to figure out who to ask than I would have wasted asking multiple people (including other people’s time I would have wasted, not just mine).

If there were a way to quickly filter out people who have added tons of wrong leading IDs that nobody has critiqued yet, or to quickly find people who are expert but haven’t done much on iNaturalist yet, then I would certainly do that. Do let us know if you have any ideas about how it could be done. For now, Pisum’s creation is the quickest way I know of to go from knowing nothing about an identifier to knowing a few useful facts about their past behaviour.

2 Likes

It’s so far beyond “suspicion” at this point though.

3 Likes

Re: the influx of Hong Kong photos. It can be frustrating dealing with so many cultivated plants, but this is a highly urbanized area. A scraggly maize plant growing in a pot* is probably as natural a sight as some of these kids have seen.

  • Two observations, same individual plant, taken from slightly different angles and posted by two different students.
4 Likes

As someone from Hong Kong, it is honestly quite frustrating to see all these cultivated plants and multiple photos of the same plant in different angles from different users. But I suppose that’s a catch-22 problem, I believe its important to get children to be more interested in the natural world and I certainly don’t want to discourage them from using the site (though in reality it’s likely once the project ends they won’t be using inat again).

I also have to point out that only about 30% of Hong Kong’s land area is urbanised. The rest of it is all natural. But I agree if the students restrict themselves within the urban HK I don’t expect much.

5 Likes