firm can now take a list of xr.datarrays for the thresholds #661
Conversation
| @@ -216,8 +219,10 @@ def _single_category_score( | |||
| # Bring back NaNs | |||
| condition1 = condition1.where(~np.isnan(fcst)) | |||
| condition1 = condition1.where(~np.isnan(obs)) | |||
| condition1 = condition1.where(~np.isnan(categorical_threshold)) | |||
There was a problem hiding this comment.
Just a note to check this extra line doesn't have any problems running when categorical_threshold is a sequence of floats. Presumably this would be covered by the pre-existing tests so may not actually need more tests, but it should be confirmed.
There was a problem hiding this comment.
There are tests for sequences of floats
|
This all looks good to me. @rob-taggart , it would be nice if you could either review the tests or do some manual testing to make sure the new functionality is working as expected and also make sure there are no issues, but nothing jumped out at me as being a problem in that regard. |
|
It's not required, but I could imagine a worked example showing say two stations from different climate zones with different thresholds, to make it clear to people why they might want to do this. On the other hand, notebooks can't cover absolutely everything in detail, and it's mentioned, so I would only do this if you felt motivated to do so. |
|
Thanks for doing this @nicholasloveday . Docstring looks good and tests are reasonable. Happy for this to be merged, @tennlee |
No description provided.