matplotlib: make legend keys square

Question

I am working with matplotlib and would like to change the keys in my legends to be squares instead of rectangles when I make, for example, bar plots. Is there a way to speci

Answer 1

If you want a very quick and dirty solution to get an approximate square (that may require some fine-tuning depending on your plot), you can adjust the handlelength kwarg in the legend call. Following Schorsch's solution (i.e. once you have a list of rectangular legend artists and corresponding labels):

ax.legend((rects1[0], rects2[0]), ('Men', 'Women'), handlelength=0.7)

See matplotlib legend() docs for more information.

Answer 2

Modifying handlelength globally affects the width of other markers in the legend. That solution will therefore be incompatible with, e.g. a combination of point and line patches. Instead, you can just add a square point marker to the legend using Line2D. You just have to set its associated line to zero-width:

rect1 = mlines.Line2D([], [], marker="s", markersize=30, linewidth=0, color="r")
rect2 = mlines.Line2D([], [], marker="s", markersize=30, linewidth=0, color="y")
ax.legend((rect1, rect2), ('Men', 'Women'))

Answer 3

You can define your own legend keys.

The bar plot in my answer is created using the matplotlib barchart demo. (I have removed the error bars). The matplotlib legend guide explains how to define a class to replace legend keys with ellipses. I have modified that class to use squares (by using rectangle patches).

import numpy as np
from matplotlib.legend_handler import HandlerPatch
import matplotlib.pyplot as plt
import matplotlib.patches as mpatches

# Define square (rectangular) patches
# that can be used as legend keys
# (this code is based on the legend guide example)

class HandlerSquare(HandlerPatch):
    def create_artists(self, legend, orig_handle,
                       xdescent, ydescent, width, height, fontsize, trans):
        center = xdescent + 0.5 * (width - height), ydescent
        p = mpatches.Rectangle(xy=center, width=height,
                               height=height, angle=0.0)
        self.update_prop(p, orig_handle, legend)
        p.set_transform(trans)
        return [p]      

# this example is the matplotlib barchart example:

N = 5
menMeans = (20, 35, 30, 35, 27)

ind = np.arange(N)  # the x locations for the groups
width = 0.35       # the width of the bars

fig, ax = plt.subplots()
rects1 = ax.bar(ind, menMeans, width, color='r')

womenMeans = (25, 32, 34, 20, 25)
rects2 = ax.bar(ind+width, womenMeans, width, color='y')

# add some text for labels, title and axes ticks
ax.set_ylabel('Scores')
ax.set_title('Scores by group and gender')
ax.set_xticks(ind+width)
ax.set_xticklabels( ('G1', 'G2', 'G3', 'G4', 'G5') )

# append the new patches to the legend-call:

ax.legend( (rects1[0], rects2[0]), ('Men', 'Women'), 
           handler_map={rects1[0]: HandlerSquare(), rects2[0]: HandlerSquare()})

plt.show()

Having defined the class HandlerSquare, one can now apply this to each legend entry as a third argument to the ax.legend call. Note the syntax:

handler_map={rects1[0]: HandlerSquare(), rects2[0]: HandlerSquare()}

The handler_map has to be a dictionary.

This will give you this plot: