Filter random instituion by country_code

[raffaem]

Why this returns a non-German institution, and how can I get a random German institution?

auth = Institutions().filter(country_code="DE").random()

[raffaem]

The following API call returns 1 random DE institution

https://api.openalex.org/institutions?page=1&filter=country_code:countries/de&sample=1

How can I replicate it in PyAlex?

[J535D165]

As far as I know, OpenAlex does not support using "random" in combination with filters (and doesn't raise an error on their side). For your example, use sample() https://github.com/J535D165/pyalex?tab=readme-ov-file#sample.

[raffaem]

As far as I know, OpenAlex does not support using "random" in combination with filters

Doesn't my post provide a direct counterexample?

The following API call returns 1 random DE institution

https://api.openalex.org/institutions?page=1&filter=country_code:countries/de&sample=1

Calling it multiple times return different institutions.

For example, here is the institution returned by the first call:

And here is the institution returned by another call:

Also, this webpage have an example in which "sample" is used in combination with filters:

Get 50 random works that are open access and published in 2021

https://api.openalex.org/works?filter=open_access.is_oa:true,publication_year:2021&sample=50&per-page=50

As you can see, there are both "filter" and "sample" in the url. And calling the URL multiple times return different entities as well.

This command gives a random Swedish institution:

https://api.openalex.org/institutions?sample=1&filter=country_code:SE

So I don't think your statement is technically correct.

[raffaem]

As far as I know, OpenAlex does not support using "random" in combination with filters (and doesn't raise an error on their side). For your example, use sample() https://github.com/J535D165/pyalex?tab=readme-ov-file#sample.

Sorry, I just realized I was talking about random while you are talking about sample.

What is the difference between the two?

So far I don't see any advantage of using random over sample?

[J535D165]

That's no problem. I also don't see the use case for random, to be honest. I assume random will be a bit more performant than sample, as it doesn't support filters.

Let's add some extra information to the README to warn users. You are not the only one.

[raffaem]

THanks