Write xs JOSS paper

[aaronspring]

very easy to create a CITATION.cff if the software is on zenodo (which xs is not yet): https://github.com/citation-file-format/doi2cff

xskillscore is actually used in a few studies already:

• all climpred using papers
• Risbey et al. 2021 http://www.nature.com/articles/s41467-021-23771-z
• probably more

Another option would be a JOSS paper, but that takes a bit more effort.

see https://github.com/pangeo-data/climpred

[raybellwaves]

You know how to upload to zendo? I got as far as here. I took a look at https://zenodo.org/record/5038065#.YQiYLVNKh6o

[aaronspring]

I would use the zenodo GitHub access https://guides.github.com/activities/citable-code/

[raybellwaves]

cheers. Looks like it needs a release to push

[aaronspring]

Or would anyone else be interested in writing a JOSS paper? @raybellwaves @bradyrx @dougiesquire

[bradyrx]

I think that could be a good thing to do, since it's a pretty short write-up. I could be available mostly to proof-read/contribute to an existing draft.

[dougiesquire]

I'd also be happy to help out. Perhaps it makes sense for @raybellwaves to lead (if he wants to/has time) given that it's his baby originally?

[raybellwaves]

on zenedo now. +1 on JOSS paper. I'm happy for whoever writes it to be lead author.

[aaronspring]

Who would be a co-author? Everyone in https://github.com/xarray-contrib/xskillscore/graphs/contributors?

[bradyrx]

For reference, here's the climpred JOSS paper: https://joss.theoj.org/papers/10.21105/joss.02781. That is basically at the length limit, so it's effectively a pretty short paper.

[raybellwaves]

Who would be a co-author? Everyone in https://github.com/xarray-contrib/xskillscore/graphs/contributors?

Would say the core devs: you, me, Riley, Dougie

[aaronspring]

What would be the criterion? More than one/5 feature/PRs merged?

[raybellwaves]

What would be the criterion? More than one/5 feature/PRs merged?

I think we can filter people > 1 PR (edited). So the author list can be:

Spring (assuming you write it :) ), Bell, Brady, Squire and Huang.

[aaronspring]

being > 1 PR

[aaronspring]

I started a first draft. I took the structure from our climpred paper https://joss.theoj.org/papers/10.21105/joss.02781.

https://sharelatex.gwdg.de/6916457693cyyrcvjcmqsw

Do we meet the JOSS requirements? I think so.

This is just a very very bare-bone first sketch. In terms of structure and content, do you feel something is missing or misleading? What would you add/remove? Please feel free to comment and also modify the text. (overleaf/sharelatex is anyways under version control so we could revert anything.)

@raybellwaves @bradyrx @dougiesquire @ahuang11

[raybellwaves]

Thanks @aaronspring. I think it meets the requirements. Worth adding it's domain agnostic e.g. https://xskillscore.readthedocs.io/en/stable/tabular-data.html. The accessor method allows it to be dropped in a pandas/scikit-learn workflow on tabular data. I wonder if this could be a separate section? i.e. it's use in ML competitions.
AFAICT no one has used it for tabular data on kaggle (https://www.kaggle.com/search?q=xskillscore) but i'm sure xss can play a part in the convos on https://www.kaggle.com/search?q=metrics. I know it's used widely in https://s2s-ai-challenge.github.io/

[dougiesquire]

Sorry for the long delay in responding @aaronspring, I've been on leave. What you have looks like a great start to me, but I'll take a better look over the coming days. How would you like to progress? Provide feedback here for you to add into the document? Or just add directly into your document ourselves?

[aaronspring]

just add directly into your document ourselves