[R] Bootstrapping
Bastian Offermann
bastian.offermann at googlemail.com
Tue Dec 6 18:17:29 CET 2011
Hello,
I am having some trouble setting up a bootstrapping prodecure. I create
artificial data and would like to use these to bootstrap a t-test
statistic from these. Especially I do not really get how boot uses the
indices variable or i variable. Can anybody help out? Thanks!!
b0 <- 1/2 # intercept as given
b1 <- 1 # slope as given
n <- 20 # of rand. numbers
u <- rnorm(n, mean=0, sd=1) # generate data
x <- rnorm(n, mean=0, sd=1) #
y <- b0 + b1*x + u # model
mu <- 1
data1 <- matrix(, n, 2) # Data from original draw
tval <- function(data1, i) {
data1[i,1] <- y
data1[i,2] <- x
m <- lm(data1[,1]~data1[,2])
p <- coef(m)[2]
t <- (coef(m)[2]-mu)/sqrt(vcov(m)[2,2]) # t-test
t
}
bs <- boot(data = data1, statistic = tval, R = 500, sim = "ordinary")
bs
ba <- boot.array(bs)[1:5,1:10]
ci <- boot.ci(bs, conf=0.95, index=1, type="all")
ci.bca <- ci$bca[4:5]
plot(bs, index=1)
More information about the R-help
mailing list