David Heiser <[EMAIL PROTECTED]> wrote in message
[EMAIL PROTECTED]">news:[EMAIL PROTECTED]...
> If however the integers represent records, in which the apparent number is
> like the number on a football jersey, then the problem is somewhat
> different. The problem here becomes selection of an unbiased choice of n
> records at random from a file containing N records.
If you mean sampling without replacement rather than with replacement,
then yes it is a different problem, but there is a really easy way to solve it.
In a loose pseudo-code, where the initial N items in a vector, L, and the
selected items are put into S. In a loose pseudo-code:
k:=N
for i=1:n {
j:= generate a random integer from 1:k
S(i):= L(j)
L(j):= L(k)
decrement k
}
However, I don't think this is the original poster's problem.
Glen
.
.
=================================================================
Instructions for joining and leaving this list, remarks about the
problem of INAPPROPRIATE MESSAGES, and archives are available at:
. http://jse.stat.ncsu.edu/ .
=================================================================
