CODEWITHSUNDEEP
pythonmatplotlib
perimosocordiae
perimosocordiae

Reputation: 17847

Using matplotlib's color cycle as a colormap

It's possible to set the color cycle to match an existing colormap, but this question is asking how to do the inverse: creating a qualitative colormap from the color cycle.

In my specific case, I've got a scatter plot with an associated array of integer class labels. My current plot looks like this:

x,y = np.random.normal(size=(2,250))
theta = np.arctan2(y,x)
c = np.digitize(theta, np.histogram(theta)[1])
plt.scatter(x,y,c=c)

current scatter plot

As you can see, this doesn't do a great job of distinguishing the classes cleanly. Instead, I'd like to somehow plug in a colormap that matches the current color cycle, where label i corresponds to color_cycle[i]. If I have more classes than the color cycle has colors, that's fine, it should just wrap around like normal.

Upvotes: 3

Views: 1650

Answers (1)

unutbu
unutbu

Reputation: 880717

I don't think there is a public API for obtaining the current color cycle, but by mimicking set_prop_cycle you might define get_prop_cycle this way:

rcParams = plt.matplotlib.rcParams
def get_prop_cycle():
    prop_cycler = rcParams['axes.prop_cycle']
    if prop_cycler is None and 'axes.color_cycle' in rcParams:
        clist = rcParams['axes.color_cycle']
        prop_cycler = cycler('color', clist)
    return prop_cycler

Once you have the colors in prop_cycler, you can map c to colors in the color cycle:

colors = [item['color'] for item in get_prop_cycle()]
c = np.take(colors, c, mode='wrap')

import numpy as np
import matplotlib.pyplot as plt
from matplotlib.rcsetup import cycler

np.random.seed(2016)

rcParams = plt.matplotlib.rcParams
def get_prop_cycle():
    prop_cycler = rcParams['axes.prop_cycle']
    if prop_cycler is None and 'axes.color_cycle' in rcParams:
        clist = rcParams['axes.color_cycle']
        prop_cycler = cycler('color', clist)
    return prop_cycler

fig, ax = plt.subplots(nrows=2)
x,y = np.random.normal(size=(2,250))
theta = np.arctan2(y,x)
c = np.digitize(theta, np.histogram(theta)[1])

ax[0].scatter(x,y,c=c)
ax[0].set_title('using default colormap')

colors = [item['color'] for item in get_prop_cycle()]
c = np.take(colors, c, mode='wrap')

ax[1].scatter(x,y,c=c)
ax[1].set_title('using color cycle')
plt.show()

enter image description here

Upvotes: 1

Related Questions