Engage with an unemployment algorithm
ledighedsalgoritmen.dk
ledighedsalgoritmen.dk
If you input that you expect to be unemployed for a short/long time, it returns that assumption. Only if you say you don't know...
Immigrant/descendant: High risk of long term unemployment, right away. (Edit: Apparently except non-western immigrants, based on the source code!)
Age is the next check that can classify you as high risk, I didn't look for the threshold (Edit: 56, according to the source). As long as you're below that, the deciding factor is...
How long were you employed out of the past 36 months? 12 months is apparently good. (Edit: <2 is the threshold).
It seems like most of the questions don't seem to be considered (if the visualization is to be believed) and it's a very simple "if x return foo" style algorithm, not some fancy scorer/ML.
Look for function setValues() in https://ledighedsalgoritmen.dk/p5vis.js
It's a decision tree. Decision trees falls under ML. The thresholds are learned from data.
I would imagine they had to use a decision tree in order to tell users and case workers _why_ they are at risk.
The algorithm has been heavily criticized for putting basically all non-western immigrants into the "high risc"-group, but if the job market historically hasn't been willing to hire high performing, non-western immigrants, isn't it the whole point of building an algorithm from data? To shed light on this?
It may be right most of the time if we lump everyone in the same bag, but that is not helpful at all.
Also, having the self-evaluated unemployment risk the first deciding factor seems to defeat the purpose.
The latter part is probably on the basis that non-Western immigrants mostly needed skilled worker visas, whereas descendents and 'Westerners' (which apparently maps to "EU members" in this case) don't. Though as with most of the rest of it I'd struggle to see this as a conclusion from advanced modelling and not just a stereotype kept relevant largely by not asking the descendent or 'Westerner' what type of work they'd been doing and how long they'd worked in Denmark
It is a long-standing taboo to split "non-Western" into "middle east" and "asian". Avoiding the split makes the bad numbers less noticeable.
It seems you've found the algorithm right there. Not only is this algorithm the good old mega-nested-if-test-of-hard-coded-answers for a few of the questions asked. It basically checks if you think you are at risk, then puts you "atRisk" if you are older than 56/about to retire/an immigrant/pregnant.
Sadly; that is probably enough to report relatively correct values.
Similar issue is often misattributed in other places like game engines aswell. For example Unity is often labeled as lower quality because of the correlation with bad code quality in many of the games made with it, while forgetting that this is precisely because it's so easy to get started with, more new developers choose it
Also, none of my questions mattered, by virtue of being a western immigrant and expecting to look for a job from 3-6 months I’m somehow at risk of long term unemployment.
I wonder if the underlying cause is "having savings and not needing a job to have a comfortable living." May also tie into how easily or not you can get government support. The more desperate you are the more likely you'll settle for any job as quickly as possible.
I guess you can already overwrite any existing function with your custom non-function with userscripts, but that seems quite hacky and non-portable, since most mobile browsers don't really support extensions (Firefox does, but only a small subsection)
After answering all of the questions, I came to a black and empty section, scrolling down from there gave me an option to do the test again.
translated to English
https://www-version2-dk.translate.goog/artikel/styrelse-frik...
on edit: changed where to are, not sure why I wrote where instead of were, but there we go.
Admittedly, it's also the type of model you'd expect someone working back from very limited available data to come up with, but I struggle to believe it's a particularly accurate model
The resulting model can be well be accurate in aggregate for informing policy decision tradeoffs. Just totally irrelevant for an individual's particular circumstances...
Thank you for posting this.
The ministry of research and education's page about their project to develop the job market for bscs.: https://ufm.dk/uddannelse/videregaende-uddannelse/fleksible-...
It's not like its impossible to get a job with a bachelor. Almost all companies and institutions will just automaticity look for master graduates, and you might have to educate them about the benefits of getting the same capabilities for a lower pay.
One issue you might encounter is that Danish IT companies are obsessed with hiring people with a masters, even if the job doesn't even require a B.Sc. Most of the jobs in the industry could be fulfilled by people from trade-schools or other shorter educations. Companies just seem to have this idea that they need engineers or people with a masters in computer science, even if their product is just some CRUD application.
Also how much is a masters degree even worth after a handful of years of relevant experience anyway? Not much...
It's not really relevant what they're afraid of. They should be made to be transparent, and it should be enforced when they don't meet the mark.
Similarily I've not really seen any sites which publish how their anti-spam filters work or what the treshold values are either for similar reasons. They're possible to figure out, but cost a lot more.
Maybe a better solution would be to somehow guarantee that they're consistent with their own rules, and regularly audited so that they're also compliant with the current law?
The worst scenario that seems to fall out of this is "evolving arms race between bad actors and the metrics".
To me, that doesn't seem anywhere near as bad as the (current) lack of transparency.
How quickly do you believe you will get a job: "Don't know" (I've enough savings to carry me a couple of years, no rush)
Origin: "Western immigrant"
Verdict: high risk of long-term unemployment.
If the little summary is to be believed, there are no other important factors. Changing my "Don't know" to 3 months reduced me to "low risk of long-term unemployment". So if I’m not mistaken, the main factor for their estimate for how long you will be unemployed … is your estimate on how long you’ll be unemployed?
A friend of my took a spent month, after getting hit masters degree in CS, finding the right job. That meant that he need to show up at the unemployment office. They where only able to offer him a forklift certification, which he didn't need or want. In the end they just left him alone for a month and he found the job he wanted.
Overall the Danish unemployment office is remarkably bad at finding jobs for people or even help them get better qualifications. There's an entire book written about it (based on one journalists experiences). In the book the only person, that particular office found a job for is one they hired themself.
How far is this demo from the real algorithm ? In the demo it seems they use a very simple decision tree just based on very few criterion (In order : Self-perceived ability to get a job, Origin, Age, Recent employment rate).
Does this simplistic (racist and ageist) model have been derived from real data and is in use, or is it just some kind of propaganda (from CopenHagen University?) and the real model they use is much more complex.
Random forests can be made much bigger and match data quite well provided they are big enough, but the bigger they get the harder they are to explain.
Caught in a landslide, no escape from reality.
You can say e.g. the resulting model surfaces 'ageist' rules, but isn't that just a reflection of reality? Jobs chances for the average >56,5yo are indeed pretty slim, and not just in Denmark. Should the model lie about that?
Thunderbolt and lightning very, very frightening.
I know nothing about this particular "algorithm" or how it is used, other than that it was to my knowledge meant for use in job centers - i.e. it applies to someone who is already unemployed and unable to find work on their own for whatever reason (we should not judge, every life is different). In some cases, those on social benefits are also forced to job centers despite their condition, which might negatively affect this crude "algorithm" as well.
In other words, it does not represent the job market or life in Denmark as a whole. With a 2.5% unemployment rate that should be more or less obvious.
Of course our job market has a degree of ageism, sexism and racism (with the rather racist nationalist parties being popular in some regions), unfortunate as it may be. A male, white Dane in their 20s is likely going to have an easier time getting jobs than a female foreigner in their early 30s, where the language barrier, additional paperwork for work visa and non-zero risk of maternity leave (can easily be 50 weeks, which is costly in industries where the standard is full rather than partial salary) may make a small company uncomfortable.
But while I cannot speak for industries I have never worked in, industries that we are familiar with here on HN are fully internationalized and would hire a foreigner in the blink of an eye, and such a person would is unlikely to ever discover a job centers.
TL;DR: don't worry about this crappy "algorithm".
Small correction: Foreigners from within the EU. Many of the Danish software companies have contracts with e.g. banks and government organisations, that makes them nervous about taking in people from outside the EU.
It depends very much on the company. It should be easy enough to find someone willing to hire you, but many won't.
I actually meant outside EU. I do not know of any places that would exclude foreigners due to government contracts. I see no lack of foreign hires within fintech and medical, and the usual government contract dumpsters (KMD, Netcompany, that sort) also hire foreigners. I myself work in a place that employs a lot of engineers from India on-site for example, and I know many companies like it.
Foreigners from within EU will have easier paperwork and living situation though.
If it's for the government to predict how many 'help finding employment' leaflets the government should print next month, that seems fine.
If it's to decide who is eligible for state handouts, it is unsuitable.
The "forwards" part would be to look at risk factors and how to mitigate or improve those points. Of course together with the unemployment office.
It's better to try and improve yourself without involvement by biased and agenda-ed state actors. Maybe with a coach. But not with g-men.
http://www.youtube.com/watch?v=YvsKIBSPfWw
movie ?
STAR is acronym for Styrelsen for Arbejdsmarked og Rekrutering (English: Agency for Labor Market and Recruitment).
They have a note describing the system (in Danish) at [1]. I copy/pasted a few relevant sections to Google Translate at [2].
TL;DR: Algorithm is a decision tree based on questionnaires and outcomes of about 152000 cases. The note states a precision of 70% and a recall of 32%. There is no mention of a split between modelling and test data.
[1] https://star.dk/media/12514/2020_01_31_beskrivelse_-_profila...
[2] https://translate.google.com/?sl=da&tl=en&text=Modellen%0APr...