How to select legend handles and labels from specified subplots inside of subplots?

Question

I have a subplot of subplots. The outer subplot consists of one row by two columns, and the two inner subplots each consist of four rows and four columns. Suppose I wanted the legend labels that correspond to only the first 2x2 inner subplot. How can I go about doing this? My attempt is below:

import numpy as np
import matplotlib.pyplot as plt
import matplotlib.gridspec as gridspec

outerD = dict(nrows=1, ncols=2)
innerD = dict(nrows=2, ncols=2)
D = dict(inner=innerD, outer=outerD)

def initialize_dubsub(D, figsize=None):
    """ """
    fig = plt.figure(figsize=figsize)
    outerG = gridspec.GridSpec(D['outer']['nrows'], D['outer']['ncols'], wspace=0.2, hspace=0.2, width_ratios=[5, 5])
    axes = []
    for n in range(D['inner']['nrows']):
        inner = gridspec.GridSpecFromSubplotSpec(D['inner']['nrows'], D['inner']['ncols'], subplot_spec=outerG[n], wspace=0.25, hspace=0.3, width_ratios=[10, 10], height_ratios=[2, 2])
        for m in range(D['inner']['nrows']*D['inner']['ncols']):
            ax = plt.Subplot(fig, inner[m])
            ax.plot([], [], label='{}x{}'.format(n, m))
            ax.set_xticks([])
            ax.set_yticks([])
            axes.append(ax)
            fig.add_subplot(ax)
    # handles, labels = axes[:4].get_legend_handles_labels() # first 2x2
    # fig.legend(handles=handles, labels=labels, loc='lower center')
    fig.legend(loc='lower center', ncol=4, mode='expand')
    plt.show()
    plt.close(fig)

initialize_dubsub(D)

This code will output 8 handles and 8 labels, whereas I want 4 each. I commented out the get_legend_handles_labels() method as this does not work on arrays.

I realize I can do ax.legend() but I prefer to use fig.legend(...). How can I achieve this?

You can programmatically prepend every label you do not want to show up in the legend with an underscore. See e.g. https://stackoverflow.com/a/52500802/8300135 — SpghttCd, Sep 28 '18 at 10:25

score 2 · Accepted Answer · answered Sep 28 '18 at 10:29

Rather than trying to call .get_legend_handles_labels on the array of subplots you want, you can just loop over the axes in that array, and append the handles and labels from those four subplots to a list.

For example:

import numpy as np
import matplotlib.pyplot as plt
import matplotlib.gridspec as gridspec

outerD = dict(nrows=1, ncols=2)
innerD = dict(nrows=2, ncols=2)
D = dict(inner=innerD, outer=outerD)

def initialize_dubsub(D, figsize=None):
    """ """
    fig = plt.figure(figsize=figsize)
    outerG = gridspec.GridSpec(D['outer']['nrows'], D['outer']['ncols'], wspace=0.2, hspace=0.2, width_ratios=[5, 5])
    axes = []
    for n in range(D['inner']['nrows']):
        inner = gridspec.GridSpecFromSubplotSpec(D['inner']['nrows'], D['inner']['ncols'], subplot_spec=outerG[n], wspace=0.25, hspace=0.3, width_ratios=[10, 10], height_ratios=[2, 2])
        for m in range(D['inner']['nrows']*D['inner']['ncols']):
            ax = plt.Subplot(fig, inner[m])
            ax.plot([], [], label='{}x{}'.format(n, m))
            ax.set_xticks([])
            ax.set_yticks([])
            axes.append(ax)
            fig.add_subplot(ax)
    handles, labels = [], []
    for ax in axes[:4]:
        handles_, labels_ = ax.get_legend_handles_labels()
        handles += handles_
        labels += labels_
    fig.legend(handles=handles, labels=labels, loc='lower center')
    #fig.legend(loc='lower center', ncol=4, mode='expand')
    plt.show()
    plt.close(fig)

initialize_dubsub(D)

score 1 · Answer 2 · answered Sep 28 '18 at 10:31

1

Try replacing

ax.plot([], [], label='{}x{}'.format(n, m))

by

ax.plot([], [], label=('' if n==0 else '_') + '{}x{}'.format(n, m))

if I understand your setup correct...

answered Sep 28 '18 at 10:31

SpghttCd

10,510
2
20
25

How to select legend handles and labels from specified subplots inside of subplots?

2 Answers2