sharex, sharey, factorplot

I found a very small bug when using factorplot.

sharex, sharey don’t get propagated to FacetGrid, line 835 in linearmodels.py.

[mpschr]

Hi @mwaskom ,

Let me say first, that I have not taken a good look at the code, so I don’t know how the interior works. In any case, when one specifies a factor as x-axis, then I guess it can be expected that sharex is directly related to the x parameter - even tough it being a factor. I don’t know if I explained well enough, but I expected the same would happen, as if x were numeric: the axis range is adapted to the data represented in the subplot.

I don’t know if you agree and think it might be implemented? If not, I’d introduce a warning message saying something the like sharex=False cannot be applied to a factor-axis.

I’ll go have a try with FacetGrid, thanks for the suggestion.

Michael

[christian-oreilly]

Same issue here. Using

g = sns.factorplot(x="bins", y="density", hue="species2", col="small_mtype", 
                   data=data, 
                   sharex=False, sharey=False,
                   rows="small_mtype",
                   size=5, linewidth=1)

I can obtain a nice plot for data which are histogram-like (i.e., normalized count-per-bin; 30 bins with range [0, 200]):

However, the range for the left plot should definitely be larger. So I would like to have 30 bins with a range [0, 400] for the left plot and [0, 200] for the right plot. Plotting such a dataset now does not give a nice result:

This is because the x-factor is shared between the other grouping factors (e.g., grid column and row factors). I wonder if it would be possible to add a switch that would allow to disable explicitly this sharing of the x/y factors across column/rows.