Use nested circles as legend - python-3.x

Question

I would like to use nested circles as legend in Python in a seaborn scatterplot where I used their size to indicate a quantity.
For now I managed, using the legend's labels and handles, to get two circles for the two extremes. Would you know how I can nest them?
I have in mind something like this:

Answer 1

I tried to draw stacked circles in legend by writing Handlers. Here is the implementation.
This will give the figure like output image and would be modified to what you need.
import matplotlib.pyplot as plt
from matplotlib.lines import Line2D
from matplotlib.collections import PathCollection
from matplotlib.legend_handler import HandlerLine2D, HandlerPathCollection
class HandlerPath(HandlerPathCollection):
"""
Custom Handler for HandlerPathCollection instances.
"""
def create_artists(self, legend, orig_handle,
xdescent, ydescent, width, height, fontsize, trans):
xdata, xdata_marker = self.get_xdata(legend, xdescent, ydescent,
width, height, fontsize)
leglines = []
legline_1 = Line2D([10], [5], marker='o', markerfacecolor='darkred', markersize=10, alpha=0.3)
legline_2 = Line2D([10], [5], marker='o', markerfacecolor='darkred', markersize=20, alpha=0.3)
legline_1.set_color("darkred")
legline_2.set_color("darkred")
leglines.append(legline_1)
leglines.append(legline_2)
return leglines
fig, ax = plt.subplots()
l = ax.plot([0.5, 0, -0.5], [0.5, 0, -0.5], linestyle = '--', color='darkred', marker='o', label="TinyTL")
c = ax.scatter([0.5, 0, -0.5], [0.5, 0, -0.5] , s = 1e3, alpha=0.3, color="darkred")
ax.legend(
l+[c], ["line", "circles"],
handler_map={
Line2D: HandlerLine2D(),
PathCollection: HandlerPath()
},
handlelength=2.5, handleheight=3
)
plt.savefig('test.png')
Hope this helps.

Answer 2

Recently, I want to achieve the same legend of the nested circles. The following is my implementation:
import matplotlib.pyplot as plt
import matplotlib.legend_handler as mhandler
import pandas as pd
# Sample data
data = [[ 0, 2, 0, 0, 0, 0, 0, 0],
[ 0, 0, 0, 0, 2, 0, 0, 0],
[ 0, 0, 0, 0, 0, 0, 2, 0],
[ 2, 0, 6, 0, 0, 6, 0, 0],
[ 2, 2, 2, 2, 3, 0, 0, 2],
[ 8, 0, 8, 0, 0, 0, 0, 0],
[ 6, 6, 6, 2, 0, 0, 0, 0],
[10, 6, 2, 0, 0, 2, 0, 0],
[10, 10, 2, 0, 4, 2, 0, 0],
[ 8, 9, 8, 20, 10, 0, 8, 2]]
df = pd.DataFrame(data).reset_index().melt(id_vars=['index'])
fig, ax = plt.subplots(figsize=(7, 7), dpi=100)
bubbles = ax.scatter(
df['index'], df['variable'],
s=df['value']*70, # So that the marker is larger
marker='o',
c='#FFB24D', edgecolor='#FF6A1C', lw=1.5,
)
# Set markers' properties before they are used as the legend handles.
def set_marker_color(handle, orig_handle):
handle.update_from(orig_handle)
handle.set_facecolor('w')
handle.set_edgecolor('k')
legend = ax.legend([bubbles], ['value'], handletextpad=2,
scatterpoints=3, # Choose three marker points for a legend entry
handler_map={
type(bubbles):
mhandler.HandlerPathCollection(
sizes=[30*70, 10*70, 2*70], # Choose the corresponding size.
marker_pad=1, # So that all points have the same x coordinate.
yoffsets=[0, -1.2, -2.1], # Offset in the y direction to get the expected layout.
update_func=set_marker_color)}, # If not, the legend handles will be same as the original handles.
frameon=False,
loc='lower left',
bbox_to_anchor=(0.01, 1.05))
After that, you can use ax.text or ax.annotate to add some labels in the legend handles.
However, I haven't figured out any way to add the number labels automatically, or to get the marker_pad and yoffsets automatically.
I hope this can be a start, and someone can find a more generic way to achieve this kind of legend.