[R] Determination of % of misclassification

[spime]

Hi R-users,

Suppose i have a two class discrimination problem and i am using logistic
regression for the classification.

> model.logit <-
> glm(formula=RES~NUM01+NUM02+NUM03+NUM04,family=binomial(link=logit),data=train.data)
> predict.logit<-predict.glm(model.logit,newdata=test.data,type='response',se.fit=FALSE)
> predict.logit

I have two questions:

1.  Suppose our training data consists of 700 observations and testing set
of 300. How can i determine no of misclassifications from predicted values
and fitted values.

2. How to determine AUC from ROC curve and also threshold value?

Waiting for reply,

Thanks in advance,

bye
-- 
View this message in context: http://www.nabble.com/Determination-of---of-misclassification-tf3899598.html#a11055026
Sent from the R help mailing list archive at Nabble.com.