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

[Peter Skomoroch]

I followed some of the discussion around datasets in the previous threads.
As you mentioned, it might make sense to make some larger datasets available
separately or as cran-style optional installs, but I think it would also be
a big plus for scipy to have some more smaller built-in datasets of various
types.

I'm thinking of something along the lines of matlab, where a large number of
demos and help are bundled with the software and the datasets used in them
are built right in.  If you go through the standard "intro to matlab" books
that come with the software, it uses these same datasets over and over.
Things like load(magic) for a magic square, sunspot data etc.  I think this
is preferable to hunting down datafiles or packages for small test cases.

Scipy already takes a similar approach in a few cases, for example:

>>> from scipy import *
>>> lena()
array([[162, 162, 162, ..., 170, 155, 128],
       [162, 162, 162, ..., 170, 155, 128],
       [162, 162, 162, ..., 170, 155, 128],
       ...,
       [ 43,  43,  50, ..., 104, 100,  98],
       [ 44,  44,  55, ..., 104, 105, 108],
       [ 44,  44,  55, ..., 104, 105, 108]])

This kind of thing just makes life a lot easier,

my .02 cents ...

-Pete


On 5/14/07, David Cournapeau <david at ar.media.kyoto-u.ac.jp> wrote:
>
> Peter Skomoroch wrote:
> > David,
> >
> > I would probably use the code quite a bit.  For me an exploration GUI
> > isn't important, but correctness and performance are.  Readability of
> > the code next after that.  Let me know if you want me to be a guinea
> > pig at some point on the algorithm correctness, I can toss some
> > additional datasets at it.
> >
> Datasets would be great ! I am converting the basic datasets available
> in R to be usable in python, but I cannot spend too much time on this,
>
> David
> _______________________________________________
> Scipy-dev mailing list
> Scipy-dev at scipy.org
> http://projects.scipy.org/mailman/listinfo/scipy-dev
>



-- 
Peter N. Skomoroch
peter.skomoroch at gmail.com
http://www.datawrangling.com
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/scipy-dev/attachments/20070514/6d955a95/attachment.html>