| Re: Random sample selection without replacement (using IMSL_RANDOM_SAMPLE) [message #78013] |
Fri, 21 October 2011 09:14  |
Michael Galloy
Messages: 1114
Registered: April 2006
|
Senior Member |
|
|
On 10/21/11 3:59 AM, Patrick Leinenkugel wrote:
> Hello,
> I like to generate a random selection (eg. 5 samples) of a population
> without replacement from a one dimensional array with n samples (eg.
> n=100). I tried to use the IMSL_RANDOM_SAMPLE function which works
> fine if I have (k,n) dimensional arrays (k variables (e.g k=2), n
> samples( e.g n=100)).
> For one dimensional arrays, however, the command
>
> samp = IMSL_RANDOM_SAMPLE(5, pop)
>
> results in the error message: "IMSL Error: IMSL_RANDOM_SAMPLE:
> Terminal error: STAT_BAD_NROW_NSAMP: NROW = 1 and NSAMP = 5. Since
> there is only one invocation of this function, NROW must be greater
> than or equal to nsamp."
>
> Can anyone tell me what I do wrong or has a nother way to easily
> generate random selection without replacement.
> Thanks,
> Patrick
I use my own routine, MG_SAMPLE. Here is the code:
http://docs.idldev.com/idllib/analysis/mg_sample.pro
Docs here:
http://docs.idldev.com/idllib/analysis/mg_sample.html
It works by creating a random array of size n and finding the indices of
the smallest k elements (using a HISTOGRAM approach instead of SORTing
the entire array).
Mike
--
Michael Galloy
www.michaelgalloy.com
Modern IDL, A Guide to Learning IDL: http://modernidl.idldev.com
Research Mathematician
Tech-X Corporation
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
