Hi all,
I've been having some discussions with folks about the relationship 
between astropy and astroML, two complementary python astronomy 
packages, and I wonder if some of these ideas could be useful as we 
think about the future of scikit-learn.

In particular, astropy has the concept of "affiliated packages", which 
are listed on their website and have some level of enforced 
compatibility with the core astropy packages (see their list here: 
http://affiliated.astropy.org/ ).  They have a well-defined set of 
criteria for packages seeking affiliation.  One nice thing they've done 
is to create a template for the code and documentation, so that any 
package which wants to seek affiliation can provide a uniform code and 
documentation style (see the template here: 
https://github.com/astropy/package-template ).

This may be an idea we could use in the future to encourage development 
of algorithms which may be too specialized or otherwise not fit well 
within scikit-learn.  In some ways, it would lead to a hierarchical 
relationship similar to that between the various scikits and the core 
scipy package.  Just as we've seen useful features from the scikits 
moved upstream into scipy, we might end up with useful development in 
affiliated packages moved upstream into scikit-learn.

Just food for thought - we can probably shelve the idea for the time 
being, but I anticipate it may be a useful concept sometime in the near 
future.
    Jake

------------------------------------------------------------------------------
Master Visual Studio, SharePoint, SQL, ASP.NET, C# 2012, HTML5, CSS,
MVC, Windows 8 Apps, JavaScript and much more. Keep your skills current
with LearnDevNow - 3,200 step-by-step video tutorials by Microsoft
MVPs and experts. ON SALE this month only -- learn more at:
http://p.sf.net/sfu/learnnow-d2d
_______________________________________________
Scikit-learn-general mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/scikit-learn-general

Reply via email to