Support adjusting the element size of legend

[zxdawn]

Description

Sometimes, the plotted scatter size is small or big, we need to adjust the size in the legend to make it easier to read.

Steps to reproduce

import proplot as pplt
import numpy as np

def rand_data():
    return np.random.uniform(low=0., high=1., size=(100,))

# Generate data.
x1, y1 = [rand_data() for i in range(2)]
x2, y2 = [rand_data() for i in range(2)]

fig, axs = pplt.subplots()
s1 = axs.scatter(x1, y1, marker='o', label='first', s=2, c='b')
s2 = axs.scatter(x2, y2, marker='o', label='second', s=3, c='r')
axs.legend([s1, s2], loc='r', ncols=1)

Expected behavior:
Support size=() in axs.legend().

Actual behavior:

I have to adjust the scatter size in the legend manually:

legend = axs.legend([s1, s2], loc='r', ncols=1)

legend.legendHandles[0]._sizes = [30]
legend.legendHandles[1]._sizes = [30]

Equivalent steps in matplotlib

import matplotlib.pyplot as plt
import numpy as np

def rand_data():
    return np.random.uniform(low=0., high=1., size=(100,))

# Generate data.
x1, y1 = [rand_data() for i in range(2)]
x2, y2 = [rand_data() for i in range(2)]

plt.figure()
plt.scatter(x1, y1, marker='o', label='first', s=2., c='b')
plt.scatter(x2, y2, marker='o', label='second', s=3., c='r')
# Plot legend.
lgnd = plt.legend(loc="lower right", scatterpoints=1, fontsize=10)
lgnd.legendHandles[0]._sizes = [30]
lgnd.legendHandles[1]._sizes = [30]

Proplot version

0.7.0

[lukelbd]

This is already a feature :). You can pass line properties to Axes.legend. Here's with linewidth=3:

import proplot as pplt
import numpy as np

def rand_data():
    return np.random.uniform(low=0., high=1., size=(100,))

# Generate data.
x1, y1 = [rand_data() for i in range(2)]
x2, y2 = [rand_data() for i in range(2)]

fig, axs = pplt.subplots()
s1 = axs.scatter(x1, y1, marker='o', label='first', s=2, c='b')
s2 = axs.scatter(x2, y2, marker='o', label='second', s=3, c='r')
axs.legend([s1, s2], loc='r', ncols=1, linewidth=3)

[lukelbd]

(Might require an update to the latest version)

[zxdawn]

Got it! Maybe, it's better to add the example to the legend section? (many matplotlib questions on StackOverflow discussed something like this).

[zxdawn]

I test it with a larger size and it looks strange:

import proplot as pplt
import numpy as np

def rand_data():
    return np.random.uniform(low=0., high=1., size=(100,))

# Generate data.
x1, y1 = [rand_data() for i in range(2)]
x2, y2 = [rand_data() for i in range(2)]

fig, axs = pplt.subplots()
s1 = axs.scatter(x1, y1, marker='o', label='first', s=2, c='b')
s2 = axs.scatter(x2, y2, marker='o', label='second', s=3, c='r')
axs.legend([s1, s2], loc='r', ncols=1, linewidth=6, markersize=6)

The manual settings work well for large values:

legend = axs.legend([s1, s2], loc='r', ncols=1)

legend.legendHandles[0]._sizes = [50]
legend.legendHandles[1]._sizes = [50]

[lukelbd]

Yeah noticed that too... in your example it looks like you're doing the equivalent of Collection.set_sizes() on the marker collection. In my example I'm doing the equivalent of Collection.set_linewidth() which (evidently) has some weird side-effects for this type of marker. I've added a 2-line workaround since this is probably a really common use-case for the "on-the-fly handle modification" feature. Now passing markersize=N will update scatter marker sizes without the weird side effect -- passing markersize updates the collection with set_sizes().

You can install the dev branch for this example to work: pip install git+https://github.com/lukelbd/proplot.git.

import proplot as pplt
import numpy as np

def rand_data():
    return np.random.uniform(low=0., high=1., size=(100,))

# Generate data.
x1, y1 = [rand_data() for i in range(2)]
x2, y2 = [rand_data() for i in range(2)]

fig, axs = pplt.subplots()
s1 = axs.scatter(x1, y1, marker='o', label='first', s=2, c='b')
s2 = axs.scatter(x2, y2, marker='o', label='second', s=3, c='r')
# axs.legend([s1, s2], loc='r', ncols=1)
leg = axs.legend([s1, s2], loc='r', ncols=1, markersize=40)

[zxdawn]

@lukelbd Thanks for the modification!