Grouped Bar Chart with More Than Two Levels

bokeh=0.13.0=py36_0
holoviews=1.10.0=py36_1

This doesn’t seem to be supported yet. From here, it says

Bars support stacking just like the Area element as well as grouping by a second key dimension.

The following doesn’t work as intended (or maybe I’m missing something?):

%%opts Bars.Grouped [group_index='Group'] Bars.Stacked [stack_index='Group']
from itertools import product
np.random.seed(3)
index, groups, other = ['A', 'B'], ['a', 'b'], ["x", "y"]
keys = product(index, groups, other)
bars = hv.Bars([k+(np.random.rand()*100.,) for k in keys],
               ["Other", 'Index', 'Group'], 'Count')
bars.relabel(group='Grouped') + bars.relabel(group='Stacked')

As a reference: https://stackoverflow.com/a/43999451/2359430

[MarcSkovMadsen]

@philippjfr , @jlstevens. Do you have time to guide a PR here?

[madeline-scyphers]

@philippjfr if I wanted to put in a PR to fix this a get the BarPlot to be able to group and stack at the same time, how much work do you think that would be? I’ve ran into this at least half a dozen times at work trying to use holoviews so I would love for it to work. I have a pretty good amount of experience with lower level bokeh and holoviews so I’m pretty sure I can do it.

is this something you’d be interested in? And is anyone else working on it right now?