Hello,
Thanks for your email..now i understand...and makes sense.
However, the reduction of fetaures is a whole different story..as it is we
came down from 22000 festures to 44 after applying statistical tests...
but in general..can anyone give me some good insights as to the differenec
beteen DIMENSION REDUCTION and VARIABLE SELECTION ? I though certain
technqiues were used for one and certain others for the other..however I
see them all being used for either of the 2 things....but then you have
some people saying "once you perform variable selection do not do
dimension reduction"???
I am confused...
Sincerely,
Monika Ray
***********************************************************************
The sweetest songs are those that tell of our saddest thought...
Computer Science, Washington University, St. Louis, MO, USA
***********************************************************************
On Fri, 10 Sep 2004, Martin Heusel wrote:
> On Thu, 9 Sep 2004 15:16:34 -0500, Monika Ray <[log in to unmask]> wrote:
>
> > When one applies the SVM..one also changes other parameters such as
> > lambda( or the conditioning parameter for quadratic programming) and
> > epsilon(the width of the e-tube in the case of svm regression) and the
> > different values necessary for a particular kernel to finally get the best
> > model. So does one get the optimal and unique solution with each
> > different value or is there a right value for each of these parameters
> > after which one can be guaranteed the unique and optimal solution?
>
> With the right value for each parameter you get the optimal solution
> regarding this
> parameter. For the optimal solution of the problem as a whole you need
> all right parameters
> and the right kernel. Fortunately for many problems a 'suboptimal'
> solution is sufficient apart
> from that with the limited precision of a computer you can't reach the
> optimal solution anyway .)
>
> > Regarding the microrray data- My opinion also matches yours. There are
> > many papers out there that do gene expression analysis and though they
> > always
> > mention the fact that the number of samples is less than the number of
> > features,
>
> Maybe many features a highly correlated or are not important for the
> problem, so an appropriate
> feature selection can reduce the features needed.
>
> Best wishes
>
> Martin
>
|