How to do linear regression in R.

First, you should read the help files on
the commands lm and step:

help(lm)
help(step)

Suppose you have three vectors y, x1 and x2
and you want to fit
y = b0 + b1 x1 + b2 x2 + error.
The following commands are useful.

mydata <- data.frame(y=y,x1=x1,x2=x2)
out    <- lm(y ~ x1 + x2,data = mydata)
names(out)
extractAIC(out)
s      <- summary(out)
print(s)
names(s)
par(mfrow=c(2,2))
plot(out,ask=F)


Another way to do regression is as follows:

X   <- cbind(x1,x2)
temp <- lsfit(X,y)
ls.print(temp)
names(temp)


Stepwise regression:

out      <- lm(y ~ x1 + x2,data = mydata)
forward  <- step(out,direction="forward")
backward <- step(out,direction="backward")
summary(forward)
summary(backward)