Wanted: open data suitable for a data science project!

[tillmanreuter@ecoevo.social]

@JorisMeys use #GBIF species occurrence point-data (@gbif ). Also spatial, and combinable: https://geodaten.bayern.de

[geospacedman@mastodon.social]

@JorisMeys This https://b-rowlingson.gitlab.io/n8workshop/Birds/birds.html is the sample analysis I did for a workshop - its very much done for teaching purposes rather than scientific ones, but there's probably a lot of science could come out of it.

[rlonstein@social.stonetools.tech]

@JorisMeys I always find interesting data at https://www.ipums.org/

[joacim@mastodon.melin.org]

@JorisMeys Oh, Pirate Bay is full of training data. If it's good enough (and legal enough) for Meta it should be good enough for the rest of us.

😄

[mkj@social.mkj.earth]

@JorisMeys Not sure if that's obvious enough but ECMWF offers a number of weather-related datasets. https://www.ecmwf.int/en/forecasts/datasets

[pl@cosocial.ca]

@JorisMeys have you heard of Jeffrey Epstein?

[janeadams@datavis.social]

@JorisMeys the VAST challenge would be perfect for this! https://github.com/vast-challenge

Background: https://datastori.es/data-stories-24-vast-challenge/

[jblsmith@sigmoid.social]

@JorisMeys It might be an obvious choice, but the UK Data Service hosts many datasets. A minority of the datasets are open, and they tend to be pretty clean already, but the massive scale of some of them makes them interesting.

https://datacatalogue.ukdataservice.ac.uk/

[jblsmith@sigmoid.social]

@JorisMeys For example, the 2019 World Risk Poll is a global survey of fears and attitudes towards risk. It has lots of demographic detail too so you can simulate the effect of different sampling strategies.

It's a 70 MB plaintext spreadsheet with 150,000 rows. It's large enough that in my last visit to the Apple Store, I could compare devices by manually timing how long it took to open the file. (It took 7 seconds on anything with an M4 chip; 5 seconds with an M5.)

https://datacatalogue.ukdataservice.ac.uk/studies/study/8739#details

[guillawme@fediscience.org]

@JorisMeys Students in bio-engineering might enjoy finding out facts about macromolecular structures from the PDB.

Both the RCSB PDB and PDBe portals offer APIs to query the meta-data about structures deposited in the PDB:

https://www.rcsb.org/docs/programmatic-access/web-apis-overview

https://www.ebi.ac.uk/pdbe/pdbe-rest-api

The very basics let you replicate the online dashboards on these portals, showing number of entries deposited or released per year. But the APIs give access to every piece of meta-data in there, so you can really ask sophisticated questions. I played a bit with it a while ago, see some examples here: https://guillawme.github.io/insights-from-the-pdb/

[rupertoverall@fosstodon.org]

@JorisMeys @genenetwork How big is big? For biology/genetics the GeneNetwork repository (https://genenetwork.org) has a lot of whole-transcriptome datasets. Each is not huge but some interesting analyses could also be done by combining them in creative ways. I’ve used this resource a lot for bioinformatics training.
Most of the datasets can be downloaded as flatfiles and the API returns JSON.

[loleg@hachyderm.io]

@JorisMeys as other folks chiming in with their national data portals are doing, here are 3‘100 OGD datasets from Switzerland with a JSON format: https://opendata.swiss/en/dataset?q=&res_format=JSON&sort=max%28res_latest_issued%2C+res_latest_modified%29+desc

[luismmontilla@fediscience.org]

@JorisMeys You could use Crossref metadata. Point them to api.crossref.org and explore millions of records about different scholarly content