Seaborn plot with colorbar, centered around 0

Question

I am trying to change the default behaviour of seaborn by adding a colormap (a continuous color palette) instead of using the hue argument, which creates bins from a continuous variable. I have found the following code to work, however, I would like to add one more option, to center the color bar at 0, that is 0 gets the color white, and the colors diverge from zero to negative/positive.

import numpy as np
import seaborn as sns
import matplotlib.pyplot as plt

y=np.random.normal(30,30,100)
x=np.random.uniform(0,50,100)

s=sns.scatterplot(
    y=y,
    x=x,
    hue=y,
    size=y,
    palette='RdBu',
    sizes=(50,50)
)
norm=plt.Normalize(y.min(),y.max())
sm=plt.cm.ScalarMappable(cmap="RdBu",norm=norm)
sm.set_array([])
s.get_legend().remove()
s.figure.colorbar(sm)

As can be seen from the image 0 gets a slightly reddish color, because the data is not symmetric about zero. How can I center the colormap around 0? I am completely fine with an inflated colormap from say -80 to 80 (because of the asymmetry) if the center is at 0.

try this: `norm=plt.Normalize(-y.max(),y.max())` or `norm=plt.Normalize(y.min(),-y.min())` — r-beginners, Oct 25 '21 at 12:04

Alex · Accepted Answer · 2021-10-25T12:15:35.183

9

Using the c, norm, and cmap key-word arguments which are passed through from seaborn to matplotlib.axes.Axes.scatter (used to colour the points instead of palette) and create a mcolors.TwoSlopeNorm to create the normalisation centred around zero you can generate the plot like so:

import numpy as np
import seaborn as sns
import matplotlib.pyplot as plt
import matplotlib.cm as cm
import matplotlib.colors as mcolors

fig, ax = plt.subplots()

y=np.random.normal(30,30,100)
x=np.random.uniform(0,50,100)

vcenter = 0
vmin, vmax = y.min(), y.max()
normalize = mcolors.TwoSlopeNorm(vcenter=vcenter, vmin=vmin, vmax=vmax)
colormap = cm.RdBu

s=sns.scatterplot(
    y=y,
    x=x,
    c=y,
    norm=normalize,
    cmap=colormap,
    ax=ax,
)

scalarmappaple = cm.ScalarMappable(norm=normalize, cmap=colormap)
scalarmappaple.set_array(y)
fig.colorbar(scalarmappaple)

edited Oct 25 '21 at 12:15

answered Oct 25 '21 at 12:03

Alex

6,610
3
20
38

I only get AttributeError: module 'matplotlib.colors' has no attribute 'TwoSlopeNorm' – Quastiat Apr 05 '22 at 09:04
@Quastiat There's not enough detail in your comment. If you open a question and include the code that you are trying to run we can help resolve this. If I had to guess I would say that `matplotlib` isn't installed properly. – Alex Apr 05 '22 at 09:19
Fantastic answer! – Tim de Silva Jul 13 '23 at 17:30

Seaborn plot with colorbar, centered around 0

1 Answers1

Linked