R help - Jul 2005 - Kernlab: problem with small datasets

Hi,

I found a small problem with kernlab. The problem, I think, is that the
3-fold cross-validation performed to estimate probabilities is not
class-balanced, so the classifier could find empty classes.
The following example (maybe a little forced) show the error:

data(glass)
set.seed(1)
model<-ksvm(Type~.,data=glass[c(1:2,75:76,165:166,180:181,190:191),],type="C-svc",kernel="vanilla",prob.model=TRUE)

Setting default kernel parameters
Error in indexes[[j]] : subscript out of bounds
In addition: Warning message:
Variable(s) `' constant. Cannot scale data. in: .local(x, ...)

HTH,
 Pablo

pablo.granitto at iasma.it wrote:
> Hi,
> 
> I found a small problem with kernlab. The problem, I think, is that the
> 3-fold cross-validation performed to estimate probabilities is not
> class-balanced, so the classifier could find empty classes.
> The following example (maybe a little forced) show the error:
> 
> data(glass)
> set.seed(1)
>
model<-ksvm(Type~.,data=glass[c(1:2,75:76,165:166,180:181,190:191),],type="C-svc",kernel="vanilla",prob.model=TRUE)
> 
> Setting default kernel parameters
> Error in indexes[[j]] : subscript out of bounds
> In addition: Warning message:
> Variable(s) `' constant. Cannot scale data. in: .local(x, ...)

So time to report it to the package mainteiner (CCing), if you think it 
is an insufficiency or bug.

Uwe Ligges

> HTH,
>  Pablo
> 
> ______________________________________________
> 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