Up and Down the Python Data and Web Visualization Stack | Hacker News Reader

Up and Down the Python Data and Web Visualization Stack

nbviewer.ipython.org

72 pointsgammarator17 comments

I really enjoyed seeing the various rich Javascript plots that are getting Python wrappers. Another favorite is Python-nvd3 ( http://nbviewer.ipython.org/gist/jdavidheiser/9552624 )

I may be biased, because I wrote the hacky IPython integration, but I really love the graphs produced by nvd3.

Author here- the slides that accompany this notebook are here: https://speakerdeck.com/wrobstory/up-and-down-the-python-dat...

I would also recommend running the notebook live- some of the interactivity/plots don't work via nbviewer.

what did you use for making the slides? they look great!

I actually just used Keynote, with the free version of Novecento Sans: http://www.myfonts.com/fonts/synthview/novecento/

> ggplot2 formally shifts to maintenance mode in favor of ggvis. https://twitter.com/dseverski/status/438083880248033280

Damn, haven't learned enough ggplot2 before it expires.

This was my favorite talk from PyData SV 2014 http://pydata.org/sv2014/abstracts/#194

Nice-do you know if videos are or will be released of the talks? Would like to see this one so there's more context on what's being shown. A little tough to follow what are probably the raw materials for the talk.

The fine folks at Facebook are recording all the talks with their fancy AV setup. I expect the talks will be on on http://pyvideo.org within the week.

Another place to check sometime in the future: http://vimeo.com/channels/pydata/videos

As a python data novice, question about ipython notebooks: Do people tend to edit and view them in the included web service, or is there a way to set it up to use your preferred editor while still having the notebook visualization experience?

The raw representation is a somewhat cluttered (to the untrained eye, anyway) json file - especially if you have checkpointed images which are bascially base64 encoded blobs! So as far as I know it's pretty much the web-editor or nothing at the moment. Obviously a JSON-aware plug-in for something like PyDev or Pycharm would be pretty neat, but I think we're a ways off that. See also http://nbdiff.org/ which solves the issue of comparing version-tracked notebooks.

There are IPython Notebook implementations for VI and Emacs which mirror much of the rich interface of the notebook. However, they do NOT include the Javascript support needed for these rich interfaces. There's also a QT frontend that can display graphics.

IPython is two separate pieces - a core kernel, and a frontend that communicates with it via a published protocol. Anybody can design any client they want to interact with the kernel. The notebook interface in the browser is definitely the richest, in terms of features.

Tangentially related - but if you're a vim fan and want to be able to use the full, in-browser notebook, you can easily set up vim bindings by following these instructions:

http://spaceli.wordpress.com/2013/10/04/add-vim-key-bindings... (the post was for iPython 1.0, but it seems to still work with 2.0)

This feature alone made me 10x more excited to use iPython notebook as a development environment.

For more involved explorations, I use the editor of my choice and import/reload the script as a module in the notebook.

IPython has %edit and %run "magics" (and probably some others that I don't know about) that be used for editor-based workflow. There are also some plugins for Vim and Emacs, but I don't have experience with those.

Is there anything like this for 3d graphs? I currently use Mayavi and PyQt, but there are a lot of dependencies. I would really like to port my stuff to the web.

Reply on news.ycombinator.com