如何绘制非方形 Seaborn Jointplot 或 JointGrid

Question

I am trying to plot my non-symmetric data using Seaborn's JointGrid. I can get it to use an equal aspect ratio, but then I have unwanted whitespace:

How do you remove the padding? The documentation for both jointplot and JointGrid simply say

size : numeric, optional

Size of the figure (it will be square).

I also tried going into feeding the extent kwarg to both jointplot and JointGrid, as well as ylim with no luck.

import numpy as np
import seaborn as sns
import matplotlib.pyplot as plt
x = np.random.normal(0.0, 10.0, 1000)
y = np.random.normal(0.0, 1.0, 1000)
joint = sns.jointplot(x, y)
joint.plot_marginals(sns.distplot, kde=False)
joint.ax_joint.set_aspect('equal')  # equal aspect ratio
plt.show() 

Answer 1

Stumbled upon this question looking for the answer myself. Having figured it out I thought I'd post the solution. As the jointplot code seems quite insistent on having the figure square I don't know if this is considered bad practice, but anyhow...

If we look through the jointplot code and follow it into JointGrid , the size parameter to jointplot (and equally JointGrid ) is used in the following expression:

f = plt.figure(figsize=(size, size))
# ... later on
self.fig = f

So to get a non-square JointGrid plot, simply run:

grid = sns.jointplot(...)
grid.fig.set_figwidth(6)
grid.fig.set_figheight(4)
grid.savefig("filename.png", dpi=300)

for a 6x4 figure.

Answer 2

For those who use Seaborn into a Jupyter Notebook, I suggest calling set_figwidht() and set_figheight() just after the sns.jointplot() command.

my_plot=sns.jointplot(x="K",y="errori",data=risultati , kind="scatter")
my_plot.fig.set_figwidth(13)

Jupyter Example

Answer 3

You will need to set the ylim and xlim parameters, which will limit the x and y axis to the tuple ranges you specify:

eg

sns.jointplot(x="n_estimators", y="learning_rate", data=data,
              color="#172235", height=8, ratio=10, space=0,
              ylim=(0, 1.1), xlim=(-20, 310))  # <-- this