# [R] Bootstrapping for residuals

David Winsemius dwinsemius at comcast.net
Thu Dec 1 00:23:19 CET 2011

```On Nov 30, 2011, at 11:31 AM, bubbles1990 wrote:

>
> Hi, Ive been trying to write a program for bootstrapping residuals
> in R but
> without much success.
>
> A lecturer wants to predict the performance of students in an end-of-
> year
> physics exam, y. The lecturer has the students results from a mid-term
> physics exam, x, and a mid-term biology exam, z.
> He proposes the following linear model for the end-of-year exam result
> yi = α + βxi + γzi + qi, where q is the error.
> Y is a matrix of the data and we have y=first column of the data and
> X=second 2 columns(the x & z data)
> Now I need to write a program for obtaining bootstrap estimates,

You replied but included no context, a common and not appreciated (in
both senses of that verb) failing in posts from Nabble. My question
would be: bootstrap estimate ... of what?

--
David.

> i have:
>
> x=scan(data)
> Y=matrix(x,ncol=3,byrow=T)
> y=Y[,1]
> X=Y[,2:3]
>
> ls=lsfit(X,y)
> beta=ls\$coef
>
> yest=beta[1]+beta[2]*X[,1]+beta[3]*X[,2]
> res=y-yest
>
> boot=function(X,res,beta,b)
> {
> n=24
> output=matrix(0,ncol=2,nrow=b)
> for(i in 1:b)
> {
> error=sample(res,n,replace=T)
> ystar=beta[1]+beta[2]*X[,1]+beta[3]*X[,2]+error
> ls=lsfit(X,ystar)
> output[i,]=ls\$coef
> }
> output
> }
>
> I think the first 8 lines are right but my function might be wrong?
> Any help?
>
> --
> View this message in context: http://r.789695.n4.nabble.com/Bootstrapping-for-residuals-tp4123657p4123657.html
> Sent from the R help mailing list archive at Nabble.com.
>

David Winsemius, MD
West Hartford, CT

```