Date:         Thu, 11 Mar 2004 06:12:40 -0800
Reply-To:     Dale McLerran <stringplayer_2@YAHOO.COM>
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         Dale McLerran <stringplayer_2@YAHOO.COM>
Subject:      Re: Stratified random sample using arrays
Comments: To: ben.powell@CLA.CO.UK
In-Reply-To:  <200403110943.i2B9h0K03493@listserv.cc.uga.edu>
Content-Type: text/plain; charset=us-ascii

Ben,

Why do you need an array?  Just sort by x and z.  In a subsequent
data step, keep a counter of how many observations with a specific
value of x have been observed.  As long as the number of
observations of stratum variable x are below the target number,
then output the observation.

  /* Use a format to identify the number of    */
  /* observations to sample from each stratum. */
  proc format;
    invalue strat_n
      1 = <p>
      2 = <q>
      3 = <r>;
  run;

  /* Construct random variable z. */
  data mydata;
    retain seed <seedvalue>;
    call ranuni(seed,z);
    drop seed;
  run;

  /* Sort z within x */
  proc sort data=mydata;
    by x z;
  run;

  /* Data step to select n{i} observations from each stratum */
  data mydata;
    set mydata;
    by x;
    if first.x then stratum_n=0;
    stratum_n + 1;
    if input(x, strat_n.)<=stratum_n then output;
    keep x y;
  run;


Best wishes,

Dale


--- ben.powell@CLA.CO.UK wrote:
> Dear SAS-L
>
> I have a dataset from which I want to make a stratified random
> sample. The
> dataset contains two variables (x,y) and are sorted by x. y is
> otherwise
> unique. I want to take p of x1, q of x2 and r of x3, etc.
>
> I know how to do this using Archer Gravely's proc rank method having
> first
> sorted by a random variable. *I DON'T HAVE PROC SURVEYSELECT*.
>
> What I would like to do is perform the same thing using arrays but
> cannnot
> work it out. The logic that I can't put into code goes:
>
> Where the dataset is presorted by a random variable z and also by x,
> load each value of x into an array and if the count is over a target
> number
> then delete; keep x and y.
>
> Any help gratefully received.


=====
---------------------------------------
Dale McLerran
Fred Hutchinson Cancer Research Center
mailto: dmclerra@fhcrc.org
Ph:  (206) 667-2926
Fax: (206) 667-5977
---------------------------------------

__________________________________
Do you Yahoo!?
Yahoo! Search - Find what you’re looking for faster
http://search.yahoo.com