Re: finding the centre of a cluster

On Jun 4, 4:21 pm, "Phil Sherrod" <PhilSher...@xxxxxxxxxxxxxxxxxxxxx>
wrote:
On 4-Jun-2007, Reef Fish <large_nassua_grou...@xxxxxxxxx> wrote:

Well, here's one example: One approach for constructing an RBF neural
network
for classification problems (with a categorical target variable) involves
initially creating a Gaussian kernel for each target category.

If your variable is categorical, where did the Gaussian nonsense come
from?

The category of the target variable for each row identifies which cluster the
row is associated with;

What has any ROW have to do with any clustering or classifying
problem?


there is one cluster per target category. The
predictor (independent) variables are continuous,

Clustering is an unstructured problem in which you DON'T KNOW even
what constitutes a cluster, let alone there is no such thing as a
predictor (independent) variable. What you folks do is to confuse
a regression problem with a CLUSTERING problem or classification
problem. Even if you use regression methods (as is often abused
by those who practice it) to find clusters, there is NO reason why
any of the predictor variables is Gaussian!


so the center of the cluster
is the mean value on each predictor dimension.

Balony!

That is sufficient to prove the gobbledy gook you neural network
classifiers are injecting into a problem of which you know NOTHING.

There are many areas of statistics where my knowledge is limited, but I believe
I have more experience with neural networks and decision trees than you do.

I believe your head is bigger than subject which you THINK you know
something. Since when does a decision tree has to be part of a
clustering problem. You don't know nearly as much about decision
trees or any STATISTICS that goes with decision trees than what you
have revealed in your posts here.

Keep dreaming.

This is the last response you'll get from me. You have to LEARN
the subject and learn it properly before you speak out on it.

-- Reef Fish Bob.

.

Relevant Pages

  • Re: HELP!!!
    ... can be improved, significantly, if supervised clustering ... *If*, that is, the class labels are correct. ... The "Gold Standard" is classification by a trained experienced human ... classification methods have an accuracy significantly exceeding the ...
    (comp.soft-sys.matlab)
  • Re: assumption of Classification
    ... Most clustering algorithms (k-means, ... > A classification tree that tries to break at every value ... normality is not the only assumption to be checked. ...
    (sci.stat.edu)
  • Re: HELP!!!
    ... I need to perform some classification on the first raw of ... and perform the classification. ... clustering a mixture of multiple class data ... corresponding MATLAB code is readily available. ...
    (comp.soft-sys.matlab)
  • Re: Prediction in clustering
    ... methods will be superior to clustering forced into predictive mode. ... > I wanted to compare two approaches (one is based on classification ... > techniques and the other based on clustering techniques) for a specific ... > "prediction in clustering"). ...
    (sci.stat.consult)
  • Re: finding the centre of a cluster
    ... What has any ROW have to do with any clustering or classifying ... I have more experience with neural networks and decision trees than you do. ...
    (sci.stat.math)