[Chicago] pandas' statsmodels vs sklearn -- opinions?

[Kevin Goetsch]

Hi Tanya,

At Braintree we use sklearn and Pandas. There's a few custom classes we had
to write to get pipelines to work but overall statsmodels just doesn't
offer the same depth of machine learning techniques. It does display it's
results in a much nicer form however.

I'd recommend sklearn and pandas but I doubt there will ever be agreement
across the board.

Cheers,

Kevin

On Wed, Mar 25, 2015 at 2:12 PM, Tanya Schlusser <tanya at tickel.net> wrote:

> Hi, all of the internet discussions on statsmodels vs sklearn are from
> 2013 or before. Is there a universally preferred way?
>
> For my part, pandas is kind of a heavy package and I spent a lot of my
> first few years in Python writing statistical models from scratch for
> clients who didn't want to install anything more than numpy -- so I'm
> partial to sklearn, but right now I want to conform to popular convention.
>
> Thanks!
>
> _______________________________________________
> Chicago mailing list
> Chicago at python.org
> https://mail.python.org/mailman/listinfo/chicago
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/chicago/attachments/20150325/f355da45/attachment.html>