[Tutor] Cluster algorithms

[Bill Mill]

Kumar,

On Wed, 26 Jan 2005 22:35:59 -0800 (PST), kumar s <ps_python at yahoo.com> wrote:
> Hi:
> 
> I am still trying to learn the OOPs side of python.
> however, things/circumstances dont seems to stop until
> I finish my practise and attaing higher understanding.
> may be, i am being pushed by circumstances into the
> stream and i am being tested if I can swim efficiently
> while I struggle with basic steps of swimming. The
> 100% analogy my perspective of learning python :-)
> 
> I have a couple of questions to ask tutors:
> 
> Are there any example programs depicting Clustering
> algorithms such as agglomerative, complete link,
> partional , squared error clustering, k-means or
> clustering algos based on Neural networks or genetic
> algorithm. although I just learned python, (to major
> extent in programming also), I need to apply some of
> these algos to my data.  Any
> suggestions/recommendations?
> 

I wrote a tutorial on using perceptrons, and gave python sample code
in the article. You can see the article at
http://www.kuro5hin.org/story/2003/11/11/17383/475 . Perceptrons
probably aren't *useful* to you in sorting your data, but the code
should provide some insight into how you can do classification in
python.

Python is excellent for data mining and classification, thanks to the
excellent numarray module as well as the ipython shell and several
visualization libraries. I use it in place of Maple, Mathematica,
Matlab, or some other specialized mathematics package.

>  Do I have to know to code well using OOP methods to
> apply these algorithms?
> 

I call some methods on matrix objects in the tutorial, but don't use
any OOP. In fact, I often find that strict OOP is not so useful for
numerical algorithms, where your code will often have to be optimized
for speed.

Peace
Bill Mill
bill.mill at gmail.com