Cross-validation in SVM

Calculation of Cross-Validation for SVM, with thoese time series which
include negative and positive values ( for example return of a stock
exchange index) must be different from a calculation of Cross-Validation
with time series which includes just absolute values( for example a
stock exchange index).

Not necessarily, depends on the type of data.

How is it calculated for a return time series?

From the man page of svm():

cross: if a integer value k>0 is specified, a k-fold cross
          validation on the training data is performed to assess the
          quality of the model: the accuracy rate for classification
          and the Mean Squared Error for regression

i.e., MSE will be used.
Z

Dear R users,

Does anyone know a solution for the problem when there are 
too many ones or zeros in the respons of a binomial model? 
I think this means that the data are over/under despersed 
and the result is very bad model fit.

I'm using glmmPQL(family=quasibinomial) to fit a model to 
my data, but the model estimates are not in the range they 
should be due to overdispersion (or under?) What shall I 
do? Is there a model type for this kind of data? I would 
prefer to keep all may data, otherwise I could also select 
some of the ones so that their number will be equal to the 
number of zeros. But I don't thik this is the right way...

Any help would be appreciated.

Thanks,
Istvan

I take it that a zero inflated negative binomial (i.e. Poisson) regression
model is what you are trying to fit, aka ZIP? If so try looking at the
documentation for the zicounts package for R, for one. Of course, you can
also search on these keywords yourself, to find exactly what you want....

Regards,

Mike

-----Original Message-----
From: r-help-bounces at stat.math.ethz.ch
[mailto:r-help-bounces at stat.math.ethz.ch] On Behalf Of I.Szentirmai
Sent: 23 February 2006 17:24
To: R-help at stat.math.ethz.ch
Subject: [R] binomial models with too many 1s???

Dear R users,

Does anyone know a solution for the problem when there are too many ones or
zeros in the respons of a binomial model? 
I think this means that the data are over/under despersed and the result is
very bad model fit.

I'm using glmmPQL(family=quasibinomial) to fit a model to my data, but the
model estimates are not in the range they should be due to overdispersion
(or under?) What shall I do? Is there a model type for this kind of data? I
would prefer to keep all may data, otherwise I could also select some of the
ones so that their number will be equal to the number of zeros. But I don't
thik this is the right way...

Any help would be appreciated.

Thanks,
Istvan

______________________________________________
R-help at stat.math.ethz.ch mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide!
http://www.R-project.org/posting-guide.html