New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Fix Probabilistic metrics #232
Conversation
Is there any reason this wasn't already in the package? Just something we missed? Anyways, thanks for adding this @aaronspring , it's an important one. It would be good to have some demos in this PR thread of it in use, since we haven't really tested the probabilistic ones too heavily in my view. |
I left it out because of this additional regarding testing of the brier score: I took it from properscoring where it is tested. |
wait until https://github.com/raybellwaves/xskillscore/issues are done |
Haven't looked at this yet, but make sure to update setup/requirements/environments to require |
I get reasonable results (reproducing Kadow et al. 2016) for |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
added some more comments
I always forget to check the Changed files tab. now resolved all. please go over them and unresolve in case you think a sections needs a modification This PR is now ready. the only metric I dont fully understand the results of is LESS. should I add a note of caution there? I am quite confident for the rest. |
Okay @aaronspring, I added a few more comments. Mostly about docstrings and stuff just to polish it up. But we should be good to merge this after you finish.
If you don't fully understand the results/are confident in them, I think you should remove LESS entirely from the package and docs. Then open an issue on implementing LESS and we can address it as its own small PR. I don't want any metrics in here that we aren't confident in, since people will be using them for their analyses Also, can you explain once more the reason why we need or want the I'll merge and update the version number after this is all done. Thanks a lot!! |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
added one small comment to testing..
reasoning for the
|
implemented all. please use the Squash and merge button. |
Thanks @aaronspring so much for all the work on this!! Will merge after these tests run, then will roll out v1.1 |
Description
threshold_brier_score
andbrier_score
from xskillscore, correctly implementcrps
andcrpss
https://github.com/raybellwaves/xskillscore/pull/20numba
added to envcompute
functions can now take kwargs (to provide threshold, comparison, ...) and metrics check themselves in kwargs for needed parameterscrpss
not assuming gaussian distribution skipped for time reasons but workscompute_pm
andcompute_hindcast
take argdim
to assign dimension to calculate metric over [cannot work for ACC]) https://github.com/bradyrx/climpred/issues/187small fixes:
pytest
lasts now shorter https://github.com/bradyrx/climpred/issues/122Fixes https://github.com/bradyrx/climpred/issues/122 https://github.com/bradyrx/climpred/issues/22 https://github.com/bradyrx/climpred/issues/218 https://github.com/bradyrx/climpred/issues/187 https://github.com/bradyrx/climpred/issues/233
Type of change
Please delete options that are not relevant.
How Has This Been Tested?
properscoring
Checklist (while developing)
pytest
, if necessary.Pre-Merge Checklist (final steps)
make html
on the documents to make sure example notebooks still compile.References
Please add any references to manuscripts, textbooks, etc.