R help - Jul 2013 - Missing data problem and ROC curves

Hello all,

Trying to get this piece of code to work on my data set. It is from
http://www.itc.nl/personal/rossiter.

logit.roc <- function(model, steps=100)
		{
		field.name <- attr(attr(terms(formula(model)), "factors"),
"dimnames")[[1]][1]
		eval(parse(text=paste("tmp <- ", ifelse(class(model$data) ==
"data.frame",
"model$data$", ""), field.name, sep="")))
		r <- data.frame(pts = seq(0, 1-(1/steps), by=1/steps), sens = 0, spec=0);
for (i in 0:steps)
		{
		      thresh <- i/steps;
		      r$sens[i] <- sum((fitted(model) >= thresh) & tmp)/sum(tmp);
		      r$spec[i] <- sum((fitted(model) < thresh) & !tmp)/sum(!tmp)
		}
		return(r)}

where model is the output of a glm.

The problem is the "sum((fitted(model) >= thresh) & tmp)" bit.
The lengths
of fitted(model) and tmp are not equal because some of the cases were
deleted from the model due to missing data! fitted(model) is a set of named
numbers while tmp is a set of integers.

My question is:
- How do I determine which cases were deleted from the model and then delete
the associated cases from tmp?

I hope this makes sense and would really appreciate any help that people may
have.

Thanks,
Tim



--
View this message in context:
http://r.789695.n4.nabble.com/Missing-data-problem-and-ROC-curves-tp4670661.html
Sent from the R help mailing list archive at Nabble.com.

On Jul 1, 2013, at 10:57 AM, tfj24 wrote:
> Hello all,
> 
> Trying to get this piece of code to work on my data set. It is from
> http://www.itc.nl/personal/rossiter.
> 
> logit.roc <- function(model, steps=100)
> 		{
> 		field.name <- attr(attr(terms(formula(model)), "factors"),
> "dimnames")[[1]][1]
> 		eval(parse(text=paste("tmp <- ", ifelse(class(model$data) ==
"data.frame",
> "model$data$", ""), field.name, sep="")))
> 		r <- data.frame(pts = seq(0, 1-(1/steps), by=1/steps), sens = 0,
spec=0);
> for (i in 0:steps)
> 		{
> 		      thresh <- i/steps;
> 		      r$sens[i] <- sum((fitted(model) >= thresh) &
tmp)/sum(tmp);
> 		      r$spec[i] <- sum((fitted(model) < thresh) &
!tmp)/sum(!tmp)
> 		}
> 		return(r)}
> 
> where model is the output of a glm.
> 
> The problem is the "sum((fitted(model) >= thresh) & tmp)"
bit. The lengths
> of fitted(model) and tmp are not equal because some of the cases were
> deleted from the model due to missing data! fitted(model) is a set of named
> numbers while tmp is a set of integers.
> 
> My question is:
> - How do I determine which cases were deleted from the model and then
delete
> the associated cases from tmp?
> 
?glm
model$na.action

-- 
David.
> I hope this makes sense and would really appreciate any help that people
may
> have.
> 
> Thanks,
> Tim
> 
> 
> 
> --
> View this message in context:
http://r.789695.n4.nabble.com/Missing-data-problem-and-ROC-curves-tp4670661.html
> Sent from the R help mailing list archive at Nabble.com.
> 
> ______________________________________________
> R-help at r-project.org 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.
David Winsemius
Alameda, CA, USA