Set up the size of backgound colour plot to the size of axes in seaborn jointplot

Question

I am using seaborn for create plot according to this example.

import numpy as np
import pandas as pd
import seaborn as sns
sns.set(style="white")

rs = np.random.RandomState(5)
mean = [0, 0]
cov = [(1, .5), (.5, 1)]
x1, x2 = rs.multivariate_normal(mean, cov, 500).T
x1 = pd.Series(x1, name="$X_1$")
x2 = pd.Series(x2, name="$X_2$")

g = sns.jointplot(x1, x2, kind="kde", size=7, space=0)

However, when I change the last line of code to

g = sns.jointplot(x1, x2, kind="kde", size=7, space=0, xlim=(-5,5), ylim=(-5,5))

the background color does not change correctly:

How I can fix the background color so that it fills the whole plot?

Answer 1

You'll need to tell the underlying function (kdeplot) to extend its KDE estimate farther out. This is accomplished through the cut argument, which is a function of the KDE bandwidth. It defaults to 3, and there's no obvious way to tell exactly how you need to set it, but it shouldn't be that hard to play around and find values that work. When using jointplot, you'll want to pass this in the joint_kws dictionary so that it gets sent to the appropriate plotting function.

sns.jointplot(x1, x2, kind="kde", size=7, space=0,
              joint_kws={"cut": 10},
              xlim=(-5,5), ylim=(-5,5))

Voila: