Add alternatives to '%matplotlib inline' in numpy lesson

[jdidion]

In newer versions of python, 'inline' appears not to be a valid backend. Two alternatives are:

1. Start ipython with the '--pylab inline' option
2. On OSX, use '%matplotlib osx' instead

[tbekolay]

Can you point me to a resource that talks about the deprecation of %matplotlib inline? My understanding is the opposite, that %matplotlib inline is the correct choice moving forward, while --pylab inline is deprecated.

Perhaps this is an install issue? IPython is now part of the Jupyter ecosystem, so to install it you should be doing pip install jupyter or conda install jupyter rather than ipython.

[tbekolay]

A post on the --pylab deprecation.

[jdidion]

I am using anaconda and have installed jupyter via conda. When I do '%matplotlib inline' I get the following error:

No event loop integration for 'inline'. Supported event loops are: qt, qt4, qt5, gtk, gtk2, gtk3, tk, wx, pyglet, glut, osx

I assumed this was a python 2 vs 3 issue, but maybe it's a problem with my machine (or with osx in general)?

Have others run into this problem on their own or their students' computers?

[valentina-s]

Are you running this command from a terminal or from a jupyter notebook. I get the same error from terminal, but not from inside a notebook.

[DealPete]

I'm using osx, but typing %matplotlib osx in the IPython terminal generates an error for me.
%matplotlib inline does the same.
After playing around with this for awhile, it seems that what you want is just %matplotlib without the inline. If it works the same on other systems, lesson #1 should probably be update to reflect this.

[davidrpugh]

Am going through the 2nd edition of Python for Data Analysis now and %matplotlib notebook is used which renders and interactive plot in the notebook. This seems superior to using %matplotlib inline which renders a static plot. According to the docs %matplotlib inline is still listed as the canonical way to enable plotting in the notebook.