Matplotlib & Jupyter - Same command returns different results in different cells

Question

First Jupyter cell:

import matplotlib.pyplot as plt
import numpy as np
import matplotlib.animation as animation
%matplotlib notebook

plt.figure()
plt.plot((1,2,3))

labels = [item.get_text() for item in plt.gca().get_xticklabels()]
print("Labels are: {}".format(labels))

The ouput is (besides the chart):

Labels are: ['', '', '', '', '', '', '', '', '', '', '']

Now, in another cell, if I run the same code:

labels = [item.get_text() for item in plt.gca().get_xticklabels()]
print("Labels are: {}".format(labels))

The ouput is:

Labels are: ['', '0.00', '0.25', '0.50', '0.75', '1.00', '1.25', '1.50', '1.75', '2.00', '']

Why the same code, referencing the same Axes, returns different results?

Probably the plt object has changed when the plot is rendered. Some internal magic in matplotlib, perhaps. — Håken Lid
– Håken Lid, Commented Jul 3, 2017 at 20:04
Thanks, Haken, but it's not that. I checked the Axes ID with id(plt.gca()) and it is the same in both cells. — Gabriel Bogo
– Gabriel Bogo, Commented Jul 3, 2017 at 20:41
I was thinking about the internal state of the object. Matplotlib has many quirks. — Håken Lid
– Håken Lid, Commented Jul 3, 2017 at 22:46

JBuete · Accepted Answer · 2017-07-04 03:41:19Z

It's an interesting problem, as far as I can tell it seems to be that while you are referencing the same object in both lots of code the actual plotting that jupyter is doing isn't side-effect free.

The plot itself is generated when the code in the cell has finished running (at least with the notebook back-end), and during the generation the details such as the xtick-labels are populated. Naïvely I assume that this is because you can add additional things to the plots that will change the x and y limits of the figure and it's more efficient to only do that once when we need to see it.

Note that you can manually set the xtick-labels prior to the generation of the plot if you know what they are going to be (or what you want them to be) ahead of time:

plt.gca().set_xticklabels(list_of_labels)

However, if you do actually need the label values prior to the plot being generated you can always use

plt.gca().get_xticks()

This will return a list of the locations of the ticks and from there it's pretty easy to make the labels from that

import matplotlib.pyplot as plt
import numpy as np
import matplotlib.animation as animation
%matplotlib notebook

# generate the figure and axes objects
plt.figure()
plt.plot((1, 2, 3))  # plot the data

# now we can look at the xtick positions, and infer the labels
labels = ["{:.2f}".format(item) for item in plt.gca().get_xticks()]
print("Labels are: {}".format(labels))

imz22 · Accepted Answer · 2017-07-03 19:50:33Z

0

instead of:

%matplotlib notebook

Use:

%matplotlib inline

answered Jul 3, 2017 at 19:50

imz22

2,9782 gold badges30 silver badges20 bronze badges

1 Comment

Gabriel Bogo Over a year ago

I actually want to use the nbAgg backend (%matplotlib notebook) to refer to the same figure.

Collectives™ on Stack Overflow

Matplotlib & Jupyter - Same command returns different results in different cells

2 Answers 2

Comments

1 Comment

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

2 Answers 2

Comments

1 Comment

Your Answer

Sign up or log in

Post as a guest

Related