R help - Mar 2013 - Lars package

Hi,
 
I'm using lars package to run some regression analysis and my doubt now is
how can I predict my model to another dataset?
Let me explain a little better:
I have a dataset from which I withhold some data. With the data that wasn't
withheld, I create the model. Now, what I'm not being able to do is apply
the model back to the data that I withheld.
Any suggestions?
 
Here it goes the code:

require(lars)
x <- as.matrix(lars.input[c(1:152)])
y <- lars.input[,c(149)]
larsdata <- as.data.frame(cbind(y,x))
attach(larsdata)
lars.model <- lars(x,y)
fit <- predict.lars(lars.model,x)$fit
plot(y ~ fit[,ncol(fit)])
#so far, so good

fit2 <- predict.lars(lars.model,withhold)$fit
plot(withhold$logBA ~ fit2[,ncol(fit2)])
#this is when the mess shows up

 
Thanks!

 
_______________________________________________________________
Giovanni Giacomin
Estudante de Engenharia de Agrimensura e Cartográfica - Universidade Federal de
Viçosa
Science Without Borders Student - Geography with emphasis in GIS - University of
Montana
	[[alternative HTML version deleted]]

Try using the "elasticnet" package instead, where I believe what you
want
is part of the package functionality. If I am not mistaken, elasticnet
largely supercedes lars -- but (someone) please correct me if I'm wrong.

-- Bert

On Tue, Mar 19, 2013 at 9:40 AM, Giovanni Giacomin
<giovannigvn@yahoo.com.br
> wrote:
> Hi,
>
> I'm using lars package to run some regression analysis and my doubt now
is
> how can I predict my model to another dataset?
> Let me explain a little better:
> I have a dataset from which I withhold some data. With the data that
> wasn't withheld, I create the model. Now, what I'm not being able
to do is
> apply the model back to the data that I withheld.
> Any suggestions?
>
> Here it goes the code:
>
> require(lars)
> x <- as.matrix(lars.input[c(1:152)])
> y <- lars.input[,c(149)]
> larsdata <- as.data.frame(cbind(y,x))
> attach(larsdata)
> lars.model <- lars(x,y)
> fit <- predict.lars(lars.model,x)$fit
> plot(y ~ fit[,ncol(fit)])
> #so far, so good
>
> fit2 <- predict.lars(lars.model,withhold)$fit
> plot(withhold$logBA ~ fit2[,ncol(fit2)])
> #this is when the mess shows up
>
>
> Thanks!
>
>
> _______________________________________________________________
> Giovanni Giacomin
> Estudante de Engenharia de Agrimensura e Cartográfica - Universidade
> Federal de Viçosa
> Science Without Borders Student - Geography with emphasis in GIS -
> University of Montana
>         [[alternative HTML version deleted]]
>
>
> ______________________________________________
> R-help@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.
>
>

-- 

Bert Gunter
Genentech Nonclinical Biostatistics

Internal Contact Info:
Phone: 467-7374
Website:
http://pharmadevelopment.roche.com/index/pdb/pdb-functional-groups/pdb-biostatistics/pdb-ncb-home.htm

	[[alternative HTML version deleted]]