Clarify how to cite pandas in scientific papers

[JoKeyser]

Dear developers,
first thank you for your great work.

I'd like to reference my use of pandas in my scientific work, but have been unable to find a recommendation how to cite this.
I'm aware of https://pandas.pydata.org/talks.html, but the "papers" seem to refer to conference proceedings (slides) that lack a publisher, and seem out-dated?

Maybe something similar to SciPy's recommendations could be added?

[TomAugspurger]

The paper linked from
https://www.scipy.org/citing.html is the correct one. We could update the link from talks. That’ll be in the pandas-website repository.

[JoKeyser]

@TomAugspurger thanks for the fast response.

So the correct bibtex entry is:

@InProceedings{ mckinney-proc-scipy-2010,
  author    = { Wes McKinney },
  title     = { Data Structures for Statistical Computing in Python },
  booktitle = { Proceedings of the 9th Python in Science Conference },
  pages     = { 51 - 56 },
  year      = { 2010 },
  editor    = { St\'efan van der Walt and Jarrod Millman }
}

I guess the only issue then is/was the dead link on https://pandas.pydata.org/talks.html to http://jarrodmillman.com/scipy2010/pdfs/mckinney.pdf when it should link to http://conference.scipy.org/proceedings/scipy2010/pdfs/mckinney.pdf

An editor asked for a publisher, but then I'll just reply that there is no one with that role.

[mroeschke]

We should eventually include citation reference in the documentations probably in overview.rst.

[JoKeyser]

@mroeschke yes, I think it's helpful to include it anywhere an author would look after it. Over in the R world, most packages come with their citation command, which makes it very convenient (and thus likely) to be cited correctly.

Not sure whether such a discussion belongs here, but it seems to me that also the work since 2010 could be included into an "official" reference recommendation?

[TomAugspurger]

Some package (astropy?) was promotion a `<package>.__citation__` variable for this: http://docs.astropy.org/en/stable/changelog.html?highlight=__citation__#id58

…

On Mon, Dec 3, 2018 at 4:00 AM Johannes Keyser ***@***.***> wrote: @mroeschke <https://github.com/mroeschke> yes, I think it's helpful to include it anywhere an author would look after it. Over in the R world, most packages come with their citation command, which makes it very convenient (and thus likely) to be cited correctly. Not sure whether such a discussion belongs here, but it seems to me that also the work since 2010 could be included into an "official" reference recommendation? — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#24036 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/ABQHIksCNVtKeDgis_7RIGu28LS3R7Q2ks5u1PYjgaJpZM4Y86S2> .

[ivalaginja]

Does pandas have a DOI for native software citation, like matplotlib? See here, where I can get a bibtex entry for native software citation, depending on the version I use or a concept DOI.

[JoKeyser]

Pandas could also get a Research Resource Identifier (RRID), see e.g. matplotlib (RRID:SCR_008624). The RRIDs of the software I used were requested in a review for the European Journal of Neuroscience.

[TomAugspurger]

I don't think pandas has either of those.

[ivalaginja]

Would the pandas team consider obtaining a DOI?

Trying to establish citation principles that will give everyone due credit, as well as create traceable records of a work unit like a piece of software, the FORCE11 Software Citation Working Group published a paper presenting software citation principles. This includes the need to cite the software product directly, additionally to software papers (see section 6.2):

6.2 Software papers. Currently, and for the foreseeable future, software papers are being published and cited, in addition to software itself being published and cited, as many community norms and practices are oriented towards citation of papers. [...] the software itself should be cited on the same basis as any other research product; authors should cite the appropriate set of software products. If a software paper exists and it contains results (performance, validation, etc.) that are important to the work, then the software paper should also be cited. We believe that a request from the software authors to cite a paper should typically be respected, and the paper cited in addition to the software.

A DOI makes native software citation possible.

Since pandas is one of the more frequently used packages in many sciences, giving people the opportunity to stick to this citation recommendation might spread this best practice in the community. It would be great to see this included and move towards this as a general standard in citing software. It sure adds to the maintenance overhead of the software, but I would recommend considering it.

[TomAugspurger]

Sure. What all does that involve? With https://guides.github.com/activities/citable-code/ it seems like we can just use Zenodo and each release gets its own DOI?

…

On Tue, Sep 3, 2019 at 1:27 PM Iva Laginja ***@***.***> wrote: Would the pandas team consider obtaining a DOI? Trying to establish citation principles that will give everyone due credit, as well as create traceable records of a work unit like a piece of software, the FORCE11 Software Citation Working Group <https://www.force11.org/about> published a paper presenting software citation principles <https://www.force11.org/software-citation-principles>. This includes the need to cite the software product directly, *additionally* to software papers (see section 6.2): 6.2 *Software papers.* Currently, and for the foreseeable future, software papers are being published and cited, in addition to software itself being published and cited, as many community norms and practices are oriented towards citation of papers. [...] *the software itself should be cited on the same basis as any other research product; authors should cite the appropriate set of software products*. If a software paper exists and it contains results (performance, validation, etc.) that are important to the work, then the software paper should also be cited. We believe that a request from the software authors to cite a paper should typically be respected, and the paper cited *in addition* to the software. A DOI makes native software citation possible. Since pandas is one of the more frequently used packages in many sciences, giving people the opportunity to stick to this citation recommendation might spread this best practice in the community. It would be great to see this included and move towards this as a general standard in citing software. It sure adds to the maintenance overhead of the software, but I would recommend considering it. — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#24036?email_source=notifications&email_token=AAKAOIUHV5JZ4ENR3WTRZO3QH2UCXA5CNFSM4GHTUS3KYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOD5ZEBFQ#issuecomment-527581334>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AAKAOIWJU32ZEBHHUEBZ3ZLQH2UCXANCNFSM4GHTUS3A> .

[ivalaginja]

Yup, that's exactly it. While Zenodo is by far not the only solution, it's definitely one of the most widely used ones and the hook with GitHub is super simple.

Yes, each release will get its own DOI so that people can cite specifically the software in the state it was in when they used it. On top of that, each software package also gets a so called "Concept DOI", which will always refer to the latest release, and the user can chose which one to cite.

The last thing would then be to add an encouragement to cite the software DOI to the citation instructions given for the software. I just realized that this repo doesn't have a CITATION.txt or similar in the first place, nor are there citation instructions in the readme, so it might be worth adding that in general - adding both a citation for the software itself, as well as any papers that should be cited.

[TomAugspurger]

Cool, thanks. I'll see about adding the Zenodo GitHub integration.

One final question, (since you seem to be familiar with this 😄) If we add a CITATION.txt or a badge to the README, does it have to be updated with each release? Or is their some kind of living "master" DOI that refers to pandas, in addition to the per-release DOIs?

[ivalaginja]

Yes, that would be the "concept DOI"! You can chose to have that in the CITATION.txt and/or in the badge and then you don't have to worry about updating that :) You can find more on the DOI business here:
http://help.zenodo.org/#versioning

At one point you can also expect the user to know or at least to research what they're doing, so sticking in the "overall" DOI (i.e. Concept DOI) should be more than fine.

[TomAugspurger]

Perfect, thanks.

@jreback @jorisvandenbossche I've requested access to add the Zenodo GitHub application to pandas-dev.