hvplot has a groupby parameter that lets you pick what variables to group the output by. This results in a widget you use to select the data subset you want to plot. If groupby is not specified, hvplot often infers what variables you want to group by. For example: ds=xr.Dataset({ 'A'['x','y','z'], np.arange(3*4*5).reshape(3,4,5)), 'B'['x','y','z'],2*np.arange(3*4*5).reshape(3,4,5)), }, coords={ 'x':range(3), 'y':range(4), 'z':range(5) } ) ds.hvplot.scatter('A','B') Even though groupby is not specified, hvplot assumes you want groupby=['x','y','z']. So, this creates a widget with sliders for x, y, and z. Since only one point exists for every combination, the resulting scatter plots only have a single point, which isn't very informative! Is there a way to force hvplot to plot all values in a single plot? Put another way, is there a way to tell it to not infer a default value for the groupby parameter, and not create a widget? In the above example, this would mean producing a scatter plot of all A and B values in the dataset. Continue reading...
