Dear All,
I am still having a problem with my function:
y1<-function(n,i,s1,s2,s3,s4)
{
a<-c(0,cumsum(rexp(n-1)))
x<-{c(rep(c(sample(1:i)),n/i))
ifelse(x>(i/2),1,ifelse(x<(i/2),0,0))}
s<-matrix(c(a+s1,a+s1+s2,a+s1+s2+s3,a+s1+s2+s3+s4),ncol=4)
}
If i try to call x i don't get a vector of o's and 1's. This is unusual
because if I use this function:
x<-function(i,n)
{
x<-c(rep(c(sample(1:i)),n/i))
ifelse(x>(i/2),1,ifelse(x<(i/2),0,0))
}
I get a vector of zeros and ones. I am sure you are all familiar with block
randomisation. I want to order s to give a random sample of numbers 1:i and
repeat it n/i number of times. I then need s to replace the each element in
the vector with 0 if it is greater than i/2 and 1 if not. I am sure that
even the function that works is very inefficient and does not use good
programming skills, but this is my first ever experience with a computer
language. Eventualy i need y1 to generate a linear combination of a,s,x
according to the laird and ware model. In order to do this i need a vector
of treatment coeffecients, this is proving troublesome, please help.
_____________________________________________________________________________________
Get more from the Web. FREE MSN Explorer download : http://explorer.msn.com
|