CODEWITHSUNDEEP
rfunctionloopsmatrixlm
Witte
Witte

Reputation: 31

Runnng lm function for every row in matrices

Hello Stackoverflow Users,

I am having trouble running an lm function in R on every row in my matrices, and I would really love if I didn't have to run it manually 1000 times. So basically I have 2 matrices with rows (1:1000) and cols (1:100), and I would like to run an lm function for each row in x for the representative row in y.

My code is:

sigmaianden=1
Beta=1
set.seed(seed=20)
xi=rnorm(100)
capn=100
capm=1000
xi.1=matrix(c(rep(c(xi),each=capm)),nrow=capm,ncol=capn)
epsiloni.1=NULL
for (i in 1:100000) {
epsiloni.1[i]=rnorm(1,0,sigmaianden*xi.1[i]^2)
}
epsiloni.11=matrix(epsiloni.1,nrow=1000,ncol=100)
yi.1=Beta*xi.1+epsiloni.11

Personally I was thinking something like:

for (i in 1:1000){
model[i]=lm(yi.1[i,1:100]~xi.1[i,1:100])
}

But that doesn't even give me a result. I will be grateful for any help.

Upvotes: 2

Views: 285

Answers (1)

Julius Vainora
Julius Vainora

Reputation: 48241

You are close. lm returns a list, so that what you want to do is to put your models to another list:

models <- vector("list", 1000)
for (i in 1:1000)
  models[[i]] <- lm(yi.1[i,1:100] ~ xi.1[i,1:100])

Then you may iterate again and extract resid(models[[i]]) or coef(models[[i]]) or anything else. However, now models is quite convenient to use, so that instead of for loops you can use just e.g.

R <- sapply(models, resid)
dim(R)
[1]  100 1000

or

C <- sapply(models, coef)
dim(C)
[1]    2 1000

Upvotes: 1

Related Questions