To give a bit more context, an example code simulating an experimental measurement. ``` import numpy as np import xarray as xr from scipy.stats import norm from matplotlib import pyplot as plt

nominal x

x = xr.DataArray(np.arange(-10, 10, 0.1), dims='index', name='x')

Shift loc

loc = xr.DataArray(np.arange(-0.5, 1, 0.5), dims='loc', name='loc')

Number of experiments

exp = xr.DataArray(range(3), dims='exp', name='exp')

Add noise to x per experiment

noise = xr.DataArray(np.random.rand(len(x), len(loc)), coords={'loc': loc}, dims=['index', 'loc']) x = x + noise * 0.5

Measure

y = xr.apply_ufunc( norm.pdf, x, x['loc'], 1, input_core_dims=[['index'], [], []], output_core_dims=[['index']], vectorize=True )

Name

x.name = 'x' y.name = 'y'

Merge

data = xr.merge([x, y]) I wish to be able to use this

Plot y against x

data.plot.line(x='x', y='y', hue='loc') However, the closest I could get is with only `y` y.plot.line(x='index', hue='loc') ```

Sorry for the late reply. I have been using this function in my projects and as such it is minimum functional. However, I will try to investigate a simpler example that replicate the issue.

Lastly, perhaps you have a better idea for groupby over multi-dimension without stacking the dimensions?

Using git bisect I am able to narrow down to this PR #3807 that introduced the regression.

I suggest that in order to convince xarrsy developers to help you is to provide an example data and show what you have tried with your string encoding solution and describe applications for the method. You should check out pandas which xarrsy extends and is more widely used then xarray. Hopefully someone have a similar problem as you with pandas and you can write here how to apply their solutions.

Sounds like a technique in data science, encoding strings, which is actually number of different techniques.

I could raise an issue on scipy's end and see what they says.

Would it be possible to add exclude_dims? I have a use case where I wanted to apply a function (for example fitting) along a dimension. I programmatically create a list of dimensions to stack to create 2D dataset. Then I loop over the stacked dimension and apply the function to last dimension.

I assumed that using a with statement I have loaded the data into memory and closing the file. with xr.open_dataset(path) as ds: data = ds However this does not load into memory entirely and so lazy loading is still in effect. So using .load does what I wanted with xr.open_dataset(path) as ds: data = ds.load()