Closed lukegre closed 3 years ago
Travis tests are failing on stats.rm_poly
for the dask arrays when called with x
and y
. When y
is forced to load()
, the tests pass. Not sure why the docs build is failing.
I have not made any changes to stats
functions.
please raise a PR on the failing stats functions. new versions of xr and/or dask likely the reason
please raise a PR on the failing stats functions. new versions of xr and/or dask likely the reason
@aaronspring I feel like we recently dealt with this in climpred
? There was some break due to the new versions and I can't remember what we had to change. I think you were working on it.
@luke-gregor, if you rebase this onto master
the bug should be fixed and you should be good to go to proceed with this.
This will be fixed from https://github.com/bradyrx/esmtools/pull/105. You can rebase once I merge it when the RTD tests pass and so on. Sorry this little implementation has been such a headache! But you found a few holes with recent upstream changes. :)
Okay @luke-gregor, another rebase onto master should do the trick here!
All passed! I'll update the docs accordingly.
Thanks @luke-gregor! Sorry for the delay here. In the final stretch of writing, so haven't kept too close of an eye on Github. Merging now!
Description
Added xarray accessors for the stats functions. Added accessor as
stats
toxr.DataArray
(and notxr.Dataset
). Includes:corr, linear_slope, linregress, polyfit, rm_poly, rm_trend
Did not change the API of these functions as I originally intended - I didn't realise how extensive the tests were and this would all have to be changed 😵
Type of change
Please delete options that are not relevant.
How Has This Been Tested?
Manually. Not a complicated implementation of xarray accessors.
Checklist (while developing)