[SciPy-dev] Presentation of pymachine, a python package for machine learning

[David Cournapeau]

Dear scipy developers and users,

   As some of you may know already, my proposal for pymachine, a python 
toolbox for machine learning in python, has been accepted for the Summer 
of Code 2007. The detailed proposal is online [1], and wikified [2]. The 
proposal timeline consists of two main steps:
   - first improving existing tools related to machine learning in 
scipy, such as they become part of "official scipy" (eg all tools in 
toolbox going into main scipy namespace). This includes scipy.cluster, 
scipy.sandbox.pyem and scipy.sandbox.svm.
   - Then building from this set of toolboxes a more high level package, 
in the spirit of similar softwares, such as orange or weka [3], 
including some visualization tools for data exploration. This part of 
the code would be put in scikits (because it will require extra 
dependencies).
All development will happen in the scipy and scikits subversion 
repositories.

   Now, before starting working on it, I would like to get some feedback 
about what other people think is necessary with respect to those goals:
   - What are the requirements for a toolbox to go from the sandbox into 
the scipy namespace ?
   - For people willing to use machine learning related software in 
python/scipy, what are the main requirements/concern ? (eg Data 
exploration GUI, efficiency, readability of the algorithms, etc...)

   cheers,

   David

[1] http://www.ar.media.kyoto-u.ac.jp/members/david/fullproposal.html

[2] http://projects.scipy.org/scipy/scipy/wiki/MachineLearning

[3] orange http://magix.fri.uni-lj.si/orange/,  weka: 
http://www.cs.waikato.ac.nz/ml/weka/