avoid seaborn influencing matplotlib plots

Question

I have already found an entry that deals with a similar topic, but the suggestion doesn't work here.

How can I use seaborn without changing the matplotlib defaults?

If I missed something I am grateful for every link.

I want to create a plot with matplotlib after having created a plot with seaborn. However, the settings of seaborn seem to affect the matplotlib appearance (I realize that seaborn is an extension of matplotlib). This happens even though I clear, close the plot etc with.

    sns.reset_orig()
    plt.clf()
    plt.close()

Complete example code:

import pandas as pd
import seaborn as sns
import matplotlib.pyplot as plt
import numpy as np
# data
df = pd.DataFrame(np.array([[1, 1], [2, 2], [3, 3]]),columns=['x', 'y'])

###### seaborn plot #######
fig=sns.JointGrid(x=df['x'],
                  y=df['y'],
                  )
#fill with scatter and distribution plot
fig = fig.plot_joint(plt.scatter, color="b")                            
fig = fig.plot_marginals(sns.distplot, kde=False, color="b")

#axis labels
fig.set_axis_labels('x','y')        

#set title
plt.subplots_adjust(top=0.92)
title='some title'
fig.fig.suptitle(title)

#clear and close figure
sns.reset_orig()
plt.clf()
plt.close()        

###### matplotlib plot #######
#define data to plot
x = df['x']
y = df['y']

#create figure and plot
fig_mpl, ax = plt.subplots()
ax.plot(x,y,'.')
ax.grid(True)
ax.set_xlabel('x')
ax.set_ylabel('y')
title='some title'
ax.set_title(title)
plt.close()

The seaborn plot always looks the same: seaborn plot

But the apperance of the matplotlib plot differs. The normal one without creating a seaborn plot in front: mpl plot normal

and how it changes if using the shown code: mpl with sns in front

how do I stop this behaviour, avoid the seaborn influencing the other plots?

Answer 1

When you import seaborn the default styling is changed.

You can change the style that matplotlib applies to plots with the plt.style.use command.

To get a list of available styles you can use plt.style.available . To change back to the classic matplotlib style you'll want to use plt.style.use('classic') .