added methods from xr.Dataset

[percygautam]

Description

Related to #1066

Checklist

• Follows official PR format
• New features are properly documented (with an example if appropriate)?
• Includes new or updated tests to cover the new feature
• Code style correct (follows pylint and black guidelines)
• Changes are listed in changelog

[percygautam]

Added the methods: isel, stack, unstack, rename, rename_vars, rename_dims

[percygautam]

Tasks:

• update 'unstack' method to include dim
• add tests
• add more methods

[percygautam]

Hi all, I have created the wrapper function for adding xr.Dataset methods:

    def extend_xr_method(func):

        @functools.wraps(func)
        def wrapped(*args, **kwargs):
            _filter = kwargs.pop('filter_groups', None)
            _groups = kwargs.pop('groups', None)
            _inplace = kwargs.pop('inplace', False)
            self = args[0]
            args = list(args)

            out = self if _inplace else deepcopy(self)

            groups = self._group_names(_groups, _filter)
            for group in groups:
                xr_data = getattr(out, group)
                args[0] = xr_data
                xr_data = func(*args, **kwargs)
                setattr(out, group, xr_data)

            return None if _inplace else out

        return wrapped

Please review, so I can implement it.

[OriolAbril]

I have not reviewed everything completely yet, will get to it whenever I have time. There is one issue that needs some discussion between us though, so I'll push for a lab meeting.

There are also several comments on docs which could also be tackled on a follow-up PR if desired.

[OriolAbril]

Have not tried yet, so could be completely wrong, please check. I fear that the loop will only work for one of the ways of calling stack (first example), and not the other (second example). I would find this quite confusing. Here are the two examples:

arr = xr.DataArray( 
     np.arange(6).reshape(2, 3), 
     coords=[("x", ["a", "b"]), ("y", [0, 1, 2])]
) 
example_1= arr.stack(z=("x", "y"))
example_2 = arr.stack({"z": ("x", "y")})

Note: I think idata.sel does the same too, so we should probably decide what do want ArviZ to do automatically and if this behaviour of kwargs working automatically vs dicts not working automatically is desired or should be fixed.

[percygautam]

Yeah, you're probably right. I have done the changes for the stack. Should I also correct other methods too, or wait for the lab-meeting?

[OriolAbril]

I'd wait for lab meeting

[percygautam]

Hey, I don't remember what was decided in the lab-meeting. Is it discussed? What to do in this case?

[percygautam]

@OriolAbril What about this one?

[percygautam]

@OriolAbril multiple tests are failing for this PR, can you help?

[OriolAbril]

I don't have time to fix mpl right now so I temporally pinned the version to <3.3 until we can make ArviZ code compatible with 3.3, I think this should fix the issues and get ci to pass again. see #1305

[OriolAbril]

I think it's ready to merge, we can polish docs in next PR.