matplotlib-users Mailing List for matplotlib

On Feb 5, 2008 2:17 PM, chombee <ch...@ne...> wrote:
> Could anyone advise me or give me an example of how to prevent a legend
> from obscuring a plot, or how I could do this better:
>
> http://www.23hq.com/seanh/photo/2862125/view-large
>
> I tried moving the legend out of the way using (x,y) coords but if I
> move it out of the axes viewport it also moves out of the figure, and it
> looks silly anyway. I also tried increasing the limits of the axes
> beyond those of the data, but that looked silly too. And I tried to
> change the font size of the legend but couldn't get it to work.
You could use custom axes (rather than subplots) and a figlegend, eg
http://matplotlib.sf.net/examples/figlegend_demo.py
JDH

Could anyone advise me or give me an example of how to prevent a legend
from obscuring a plot, or how I could do this better:
http://www.23hq.com/seanh/photo/2862125/view-large
I tried moving the legend out of the way using (x,y) coords but if I
move it out of the axes viewport it also moves out of the figure, and it
looks silly anyway. I also tried increasing the limits of the axes
beyond those of the data, but that looked silly too. And I tried to
change the font size of the legend but couldn't get it to work.
I think what I need is legends that are outside of the axes viewport but
still inside the figure (the figure needs to get larger) and that are in
a smaller font.
Btw I know I should round off my statistical numbers but that isn't
going to fix it completely.
Thanks!

Carol Leger wrote:
> Hi Folks,
> 
> I have acquired some code that was running on a previous version of 
> matplotlib. I am now using version 0.91.2.
> 
> The code I acquired uses the following code fragment:
> import pylab
> from matplotlib.backends.backend_agg import FigureCanvasAgg as FigureCanvas
> <other stuff>
> figure = matplotlib.pylab.Figure( figsize=(8,6), # 8"x6" is default
> facecolor='r' )
> canvas = FigureCanvas( figure )
> 
> # ( left, bottom, width, height )
> axes_ts = figure.add_axes( ( 0.05, 0.0, 0.75, 0.04 ) )
> axes_ts.set_axis_off()
> 
> 
> subplots = []
> 
> axes_ll = figure.add_axes( ( 0.05, 0.05, 0.9, 0.9 ) ) # l, l
> subplots.append( (axes_ll, data_sets[ 0 ]) )
> 
> for subplot in subplots:
> axes = subplot[ 0 ]
> data_set = subplot[ 1 ]
> axes.set_xlim( -1.0, 1.0 )
> axes.set_ylim( -1.0, 1.0 )
> axes.set_aspect( 'equal' , fixLimits=True ) <+++++++
> axes.set_axis_off()
> data_set.render( axes, fm, to )
> 
> 
> It seems that axes.set_aspect does not have a keyword of fixLimits. Did 
> it ever? Does it depend on the backend? The original code used GTKAgg. 
> I am not using GTKAgg. I am just using a default backend.
Carol,
That kwarg was eliminated a little less than 2 years ago as part of a 
major reworking of aspect ratio handling. It never depended on the 
backend.
The kwarg equivalent to fixLimits=True is adjustable='box'. Quite a bit 
has changed in the last two years, though, so there may be other things 
in your code that need to be updated.
Eric