Keyword confusion on chunking in prediction module about climpred HOT 5 CLOSED

Yep. I will look into it. There is probably even a xr built in function that does something like chunking.

from climpred.

@aaronspring , there is chunking for xarray: http://xarray.pydata.org/en/stable/generated/xarray.Dataset.chunk.html

Which is why I was confused on our call how this chunking differs. If it is actually different than dask/xarray chunking we should probably use different language.

from climpred.

my chunking function is definately different.

the gist of what it does:

• number_chunks=True: it breaks a dataset with dimension dim into N pieces into a new dataset with dimensions dim and c.
• chunk_length=True: it breaks a dataset with dimension dim into pieces of length N into a new dataset with dimensions dim and c

Its therefore somehow similar to stack/unstack but not really. The function nicely reflects what I get from the meaning of chunking from a dictionary and from Boer 2004 paper.

Maybe I rename it chunk_Boer? Still not happy about that.

from climpred.

I think for now you are fine to leave it just as "chunking" because we are aware that it's under esmtools.prediction and not xr.chunking. Maybe the doc string can be clear that these are different operations.

This is a low-priority issue, but I'm sure we can find some smart solution through stackexchange. I think it's fine as is, but I'm curious from a programming perspective how to tackle this. It might be good for you to post a question, because I'm not finding anything really.

from climpred.

as it is only an internal function _chunking inside DPP, I think we can close this

from climpred.