What are the benefits of this over something like Bokeh?

... or matplotlib?

Does matplotlib let you put interactive visualisations in iPython notebooks?

Not really. You can use widgets and redraw when the user changes something, but this gets clumsy quickly.

Bokeh allows for interactive visualization though. Vispy[1] does to, by using webgl (but this require a running ipython kernel - so the plot will be static when exported to HTML).

This[2] is an interesting blog post about the future of python plotting/visualization.

[1] http://vispy.org/

[2] http://www.almarklein.org/future_vis.html

With matplotlib you don't need to redraw the whole plot or even the whole axes you can just change the data and update the relevant artist.

Yes, you can use the nbagg backend [1].

[1] http://matplotlib.org/users/whats_new.html#the-nbagg-backend

You can easily place the matplotlib plot in a widget and e.g. one can allow the user to control it with sliders and input boxes etc.

Looks like the main advantage is bqplot allows you to create interactive data visualizations. Bokeh and Plotly (commercial product) are the existing libraries that do this, but leave a lot to be desired.

What else are you looking for in Bokeh and Plotly?

Is there any documentation for the plotting library? I would love to explore it, but couldn't find many examples / documentation for different kinds of plot.

Interesting to note the use of IPython as a research tool even within a corporate financial firm.

It's also used heavily within Google.

I think that a lot of companies used to have internal tools like IPython, but now IPython/Jupyter has surpassed them all. Hooray for OSS.

Nice to know. I was given to understand Matlab was heavily used inside of Google, but that was a while ago.

Julia, Python and Jupyter is awesome toolkit for research, presentation and collaboration.

Hooray indeed to the awesome IPython team!