seaborn.objects: Rotation of ticks

[JanPalasek]

Hello,

might I suggest a new feature request: rotation of the ticks for seaborn objects? Or is it possible at this moment? I haven't found it yet.

Currently I use .on(axis).plot() and then on this axis I set it up via matplotlib with axis.xaxis.set_tick_params(rotation=90).

[mwaskom]

I assume you mean rotation of tick labels? I don't think that's currently supported since it's not (AFAICT) something you can set a default for with rcParams.

[JanPalasek]

I assume you mean rotation of tick labels? I don't think that's currently supported since it's not (AFAICT) something you can set a default for with rcParams.

I presume the implementation is difficult then. That's pity, data scientists from my bubble need it quite often. If there was a switch for example so.Continuous().label(rotation=90), it would be awesome.

[shoaibkhanz]

@JanPalasek quite naturally, I tried something very similar, it could also be part of the tick method fig.scale(x=so.Nominal().tick())

[mwaskom]

That's a reasonable place expect the parameter, but implementation-wise, it doesn't make sense. Scale.tick and Scale.format are interfaces to define matplotlib Locator/Formatter objects. But those can't control the label rotation, which is a property of the label artist.

[JanPalasek]

@mwaskom Could this interface be expanded?

def label(self, rotation=None,  formatter_param1=None, formatter_param2=None, formatter=None):
  pass

Or even decide on each individual parameter, whether it is a parameter for formatter or for some other component (artist?).

[Demetrio92]

seaborn.axisgrid.FacetGrid has a method .set_xticklabels with an option to rotate the labels. However, on my example it's currently buggy #3341

[GStechschulte]

See the answer by Brian Head in this StackOverflow response.

It has worked well in plots with faceting and or no-facets.

p = (
    so.Plot(data=df, x="...", y="...")
    .facet(col="...")
    .add(so.Bars())
    .theme({"figure.figsize": (16, 6)})
    ).plot() # Notice the call to plot

# If the facet includes two columns
p._figure.axes[0].xaxis.set_tick_params(rotation=90)
p._figure.axes[1].xaxis.set_tick_params(rotation=90)
p.show()