NYC Open Statistical Programming Meetup on 9/14/2011

Link to recording of talk

I had the privilege of speaking yesterday at the first meeting of the former NYC R Meetup, now dubbed the NYC Open Statistical Programming Meetup. I was really excited to share some of the new stuff I’ve built lately in pandas, though I didn’t have enough time to demo features as I would have liked :) I also ran out of time to talk about statsmodels, but I’ll post some materials here for those who are interested.

One thing I showed off which a lot of people have been asking me about is the totally awesome new IPython HTML notebook. I don’t think I’ll ever do a Python demo in anything else, it’s just that good for teaching and doing interactive demos. Brian Granger, Fernando Perez, and crew have developed a notebook file format (.ipynb) for saving and sharing these interactive notebooks. I’ll post links to my IPython notebooks for the pandas demo and the statsmodels demo I didn’t have time for at the end, along with instructions for running them.

Here are my slides from the talk:

Now, here’s the exciting part: the IPython notebooks. You’ll need to install the latest development version of IPython from GitHub. Make sure you have the Python ZeroMQ bindings and install the tornado web server. This ought to work for those (using setuptools):

easy_install pyzmq
easy_install tornado

Once you’ve done all that, fire up a terminal and run:

ipython notebook --pylab=inline

You should see the server fire up and give you a web address like 127.0.0.1:8888. Now, you need my demo workbooks. Here is a zip file with everything you need:

Demo Notebooks
PDF of pandas demo notebook
PDF of statsmodels demo notebook

Expand this zip file and launch the IPython notebook in the same directory where the .ipynb live. You should be able to run all the code in the pandas demo workbook yourself if you have pandas 0.4 (click “Run All”), but the statsmodels notebook uses some very new code from the pandas-integration branch on GitHub and some formula code that is not part of the statsmodels repo. So you can see all the code and output, but if you want to know more, I encourage you to get involved in the project!

IPython HTML Notebook