5 ms·

Up and Down the Python Data and Web Visualization Stack

72 points by gammarator 12y ago

gourneau 12y ago
This was my favorite talk from PyData SV 2014 http://pydata.org/sv2014/abstracts/#194 http://pydata.org/sv2014/abstracts/#194
- tdicola 12y ago
  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.
  - gourneau 12y ago
    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 http://pyvideo.org within the week.
    - plumeria 12y ago
      Another place to check sometime in the future: http://vimeo.com/channels/pydata/videos http://vimeo.com/channels/pydata/videos
kcorbitt 12y ago
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?
- TimSAstro 12y ago
  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/ http://nbdiff.org/ which solves the issue of comparing version-tracked notebooks.
- squishyphysics 12y ago
  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.
  - bencpeters 12y ago
    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-for-ipython-1-0-0/ 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.
- bpowah 12y ago
  For more involved explorations, I use the editor of my choice and import/reload the script as a module in the notebook.
- zokier 12y ago
  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.
squishyphysics 12y ago
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 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.
Enginoob 12y ago
Author here- the slides that accompany this notebook are here: https://speakerdeck.com/wrobstory/up-and-down-the-python-data-and-web-visualization-stack 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.
- glamp 12y ago
  what did you use for making the slides? they look great!
  - Enginoob 12y ago
    I actually just used Keynote, with the free version of Novecento Sans: http://www.myfonts.com/fonts/synthview/novecento/ http://www.myfonts.com/fonts/synthview/novecento/
    - pwang 12y ago
      <steals Rob's font...>
- est 12y ago
  > ggplot2 formally shifts to maintenance mode in favor of ggvis. https://twitter.com/dseverski/status/438083880248033280 https://twitter.com/dseverski/status/438083880248033280 Damn, haven't learned enough ggplot2 before it expires.
broken_symlink 12y ago
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.

72 points by gammarator 12y ago

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

tdicola 12y ago
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.
- gourneau 12y ago
  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 http://pyvideo.org within the week.
  - plumeria 12y ago
    Another place to check sometime in the future: http://vimeo.com/channels/pydata/videos http://vimeo.com/channels/pydata/videos

kcorbitt 12y ago

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?

TimSAstro 12y ago
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/ http://nbdiff.org/ which solves the issue of comparing version-tracked notebooks.
squishyphysics 12y ago
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.
- bencpeters 12y ago
  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-for-ipython-1-0-0/ 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.
bpowah 12y ago
For more involved explorations, I use the editor of my choice and import/reload the script as a module in the notebook.
zokier 12y ago
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.

squishyphysics 12y ago

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 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.

Enginoob 12y ago

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

glamp 12y ago
what did you use for making the slides? they look great!
- Enginoob 12y ago
  I actually just used Keynote, with the free version of Novecento Sans: http://www.myfonts.com/fonts/synthview/novecento/ http://www.myfonts.com/fonts/synthview/novecento/
  - pwang 12y ago
    <steals Rob's font...>
est 12y ago
> ggplot2 formally shifts to maintenance mode in favor of ggvis. https://twitter.com/dseverski/status/438083880248033280 https://twitter.com/dseverski/status/438083880248033280 Damn, haven't learned enough ggplot2 before it expires.

broken_symlink 12y ago

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.