python - How to make IPython notebook matplotlib plot inline

886
votes

I am trying to use IPython notebook on MacOS X with Python 2.7.2 and IPython 1.1.0.

I cannot get matplotlib graphics to show up inline.

import matplotlib
import numpy as np
import matplotlib.pyplot as plt
%matplotlib inline

I have also tried %pylab inline and the ipython command line arguments --pylab=inline but this makes no difference.

x = np.linspace(0, 3*np.pi, 500)
plt.plot(x, np.sin(x**2))
plt.title('A simple chirp')
plt.show()

Instead of inline graphics, I get this:

<matplotlib.figure.Figure at 0x110b9c450>

And matplotlib.get_backend() shows that I have the 'module://IPython.kernel.zmq.pylab.backend_inline' backend.

10
pythonmatplotlibjupyter-notebookipythondata-visualization
your code snippet should not produce <matplotlib.figure.Figure at 0x110b9c450> but <matplotlib.text.Text at 0x94f9320> (because your last line is printing a title). Anyway, your code (with %matplotlib inline and plt.show()) works as expected on windows - joaquin
Thanks for those suggestions, but they don't work for me. I still get the above output with no inline graphics. Do you have any troubleshooting advice? - Ian Fiske
no clue. Same python, same ipython (and same backend) but on windows, and it works.... I suppose plot is working for you when not inline, right ? - joaquin
without the %matplotlib inline, the kernel stays busy permanently and I get no output. It has to be killed. This is trying to use the MacOSX backend but I guess it cannot be opened for some reason. When not using ipython notebook, the MacOSX backend for matplotlib works just fine. - Ian Fiske
I had an identical symptom but it turned out I had installed a 32 bit version of Canopy on OSX 10.8. Reinstalling with the 64 bit version fixed it. - Vicky T

10 Answers

1188
votes

I used %matplotlib inline in the first cell of the notebook and it works. I think you should try:

%matplotlib inline

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

You can also always start all your IPython kernels in inline mode by default by setting the following config options in your config files:

c.IPKernelApp.matplotlib=<CaselessStrEnum>
  Default: None
  Choices: ['auto', 'gtk', 'gtk3', 'inline', 'nbagg', 'notebook', 'osx', 'qt', 'qt4', 'qt5', 'tk', 'wx']
  Configure matplotlib for interactive use with the default matplotlib backend.
225
votes

If your matplotlib version is above 1.4, it is also possible to use

IPython 3.x and above

%matplotlib notebook

import matplotlib.pyplot as plt

older versions

%matplotlib nbagg

import matplotlib.pyplot as plt

Both will activate the nbagg backend, which enables interactivity.

Example plot with the nbagg backend

96
votes

Ctrl + Enter

%matplotlib inline

Magic Line :D

See: Plotting with Matplotlib.

25
votes

Use the %pylab inline magic command.

16
votes

To make matplotlib inline by default in Jupyter (IPython 3):

  1. Edit file ~/.ipython/profile_default/ipython_config.py

  2. Add line c.InteractiveShellApp.matplotlib = 'inline'

Please note that adding this line to ipython_notebook_config.py would not work. Otherwise it works well with Jupyter and IPython 3.1.0

12
votes

I have to agree with foobarbecue (I don't have enough recs to be able to simply insert a comment under his post):

It's now recommended that python notebook isn't started wit the argument --pylab, and according to Fernando Perez (creator of ipythonnb) %matplotlib inline should be the initial notebook command.

See here: http://nbviewer.ipython.org/github/ipython/ipython/blob/1.x/examples/notebooks/Part%203%20-%20Plotting%20with%20Matplotlib.ipynb

8
votes

I found a workaround that is quite satisfactory. I installed Anaconda Python and this now works out of the box for me.

6
votes

I did the anaconda install but matplotlib is not plotting

It starts plotting when i did this

import matplotlib
import numpy as np
import matplotlib.pyplot as plt
%matplotlib inline
4
votes

I had the same problem when I was running the plotting commands in separate cells in Jupyter:

In [1]:  %matplotlib inline
         import matplotlib
         import matplotlib.pyplot as plt
         import numpy as np
In [2]:  x = np.array([1, 3, 4])
         y = np.array([1, 5, 3])
In [3]:  fig = plt.figure()
         <Figure size 432x288 with 0 Axes>                      #this might be the problem
In [4]:  ax = fig.add_subplot(1, 1, 1)
In [5]:  ax.scatter(x, y)
Out[5]:  <matplotlib.collections.PathCollection at 0x12341234>  # CAN'T SEE ANY PLOT :(
In [6]:  plt.show()                                             # STILL CAN'T SEE IT :(

The problem was solved by merging the plotting commands into a single cell:

In [1]:  %matplotlib inline
         import matplotlib
         import matplotlib.pyplot as plt
         import numpy as np
In [2]:  x = np.array([1, 3, 4])
         y = np.array([1, 5, 3])
In [3]:  fig = plt.figure()
         ax = fig.add_subplot(1, 1, 1)
         ax.scatter(x, y)
Out[3]:  <matplotlib.collections.PathCollection at 0x12341234>
         # AND HERE APPEARS THE PLOT AS DESIRED :)
3
votes

You can simulate this problem with a syntax mistake, however, %matplotlib inline won't resolve the issue.

First an example of the right way to create a plot. Everything works as expected with the imports and magic that eNord9 supplied. 

df_randNumbers1 = pd.DataFrame(np.random.randint(0,100,size=(100, 6)), columns=list('ABCDEF'))

df_randNumbers1.ix[:,["A","B"]].plot.kde()

However, by leaving the () off the end of the plot type you receive a somewhat ambiguous non-error.

Erronious code:

df_randNumbers1.ix[:,["A","B"]].plot.kde

Example error:

<bound method FramePlotMethods.kde of <pandas.tools.plotting.FramePlotMethods object at 0x000001DDAF029588>>

Other than this one line message, there is no stack trace or other obvious reason to think you made a syntax error. The plot doesn't print.