I know that I can get the positions of my y-ticks by ax.get_yticks() (btw., is that the best/correct way to get them?). But I need the tick positions relative to the axis limits (i.e., between 0 and 1). What is the best way to get that? I tried ax.get_yticks(transform=ax.transAxes), which does not work.
Asked
Active
Viewed 6,828 times
4
frixhax
- 1,325
- 3
- 18
- 30
-
I do not understand your question at all – tacaswell Jan 13 '15 at 03:45
-
2@tcaswell I think he wants to get the tick positions in Axes coordinates. – hitzg Jan 13 '15 at 08:34
-
_why_ do you want this? I have a suspicion that you are doing something the very hard way ;) – tacaswell Jan 15 '15 at 17:28
-
Because [this](http://stackoverflow.com/q/27910125/1812066) does not work and I have to get rid of overlapping labels while having no gap between the plots. – frixhax Jan 15 '15 at 17:35
1 Answers
6
This transformation can be easily done by hand:
y_min, y_max = ax.get_ylim()
ticks = [(tick - y_min)/(y_max - y_min) for tick in ax.get_yticks()]
UPDATE
To get it working with log scales, we need to employ matplotlib.transforms. But transforms can only work with (x,y) points, so we need first to complement y-ticks with some x coordinates (for example, zeros):
crd = np.vstack((np.zeros_like(ax.get_yticks()), ax.get_yticks())).T
Then we can transform data coodinates to display and then to axes and take y column of it:
ticks = ax.transAxes.inverted().transform(ax.transData.transform(crd))[:,1]
Andrey Sobolev
- 12,353
- 3
- 48
- 52
-
-
Exactly - for what I'm doing it is working. But maybe there is some more general approach? – frixhax Jan 14 '15 at 00:17
-
For more general approach you can always tinker with transforms, but it's much more cumbersome and not pythonic. Nevertheless, for the sake of completeness I'm adding it to the answer. – Andrey Sobolev Jan 14 '15 at 07:37