[R] How to find AUC in SVM (kernlab package)

[Muhammad Subianto]

On this day 11/24/2006 05:03 PM, Amir Safari wrote:
> Hi
> you need predict.ksvm() function.
>
Yes, like this is below I do to predict (kernlab package, 
http://www.jstatsoft.org/v11/i09/v11i09.pdf):
pima.pred <- predict(pimamodel, Pima.te[,-8], type="probabilities")
pima.pred

My problems is how to find AUC (with ROCR package, or other ROC 
functions) from predict above.

Regards, Muhammad Subianto

> for more information see The kernlab package here:
>  http://lib.stat.cmu.edu/R/CRAN/doc/packages/kernlab.pdf
>
> cheers,
> Amir
>
>
> */Muhammad Subianto <msubianto at gmail.com>/* wrote:
>
>     Dear all,
>     I was wondering if someone can help me. I am learning SVM for
>     classification in my research with kernlab package. I want to know
>     about
>     classification performance using Area Under Curve (AUC). I know ROCR
>     package can do this job but I found all example in ROCR package have
>     include prediction, for example, ROCR.hiv {ROCR}. My problem is
>     how to
>     produce prediction in SVM and to find AUC.
>
>     Here is a simple example:
>
>     library(MASS)
>     library(kernlab)
>     library(ROCR)
>
>     pimamodel <- ksvm(type ~
>     .,data=Pima.tr,type="C-svc",C=10,prob.model=TRUE)
>     pimamodel
>     fitted(pimamodel)
>
>     pima.pred <- predict(pimamodel, Pima.te[,-8], type="probabilities")
>     pima.pred
>
>     # try to find AUC
>     #predid.no <- prediction(pima.pred[,1], Pima.te[,8])
>     #predid.yes <- prediction(pima.pred[,2], Pima.te[,8])
>     predid <- prediction(pima.pred, Pima.te[,8])
>     perfid <- performance(predid,"tpr","fpr")
>     perfid.auc <- performance(predid,"auc")
>     perfid.auc
>
>     Thank you very much for your help.
>
>     Best wishes, Muhammad Subianto
>
>     ______________________________________________
>     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
>     and provide commented, minimal, self-contained, reproducible code.
>
>
> ------------------------------------------------------------------------
> Cheap Talk? Check out 
> <http://us.rd.yahoo.com/mail_us/taglines/postman8/*http://us.rd.yahoo.com/evt=39663/*http://voice.yahoo.com> 
> Yahoo! Messenger's low PC-to-Phone call rates.