On Fri, Nov 16, 2012 at 7:58 AM, Skipper Seabold <jsseabold@gmail.com> wrote:
On Fri, Nov 16, 2012 at 10:19 AM, Yaroslav Halchenko <sf@onerussian.com> wrote:
> I just found some code (http://www.onerussian.com/tmp/plots.py and
> pasted below for review/feedback) laying around which I wrote around
> matplotlib for plotting primarily pair-wise stats results.  Here is a
> demonstration:
> http://nbviewer.ipython.org/url/www.onerussian.com/tmp/run_plots.ipynb
> I wonder if there is a need/place for it in matplotlib and what changes would
> you advise. Sorry for the lack of documentation -- I guess I have not finished
> it at that point (scipy dependency can easily be dropped, used only for
> standard error function iirc):

Looks nice. We'd certainly be interesting in including it in
statsmodels/graphics if there isn't sufficient interest here and/or
you'd like to keep the scipy dependency. ;)


I was going to suggest either the same thing or adding it to pandas. I think statsmodels if the better fit, though. I also noticed scipy is only used for scipy.stats.sem -- so it might be easy enough to loose the scipy dependency. Just a thought.