Date:         Tue, 6 Dec 2005 23:41:15 -0500
Reply-To:     "Howard Schreier <hs AT dc-sug DOT org>" <nospam@HOWLES.COM>
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         "Howard Schreier <hs AT dc-sug DOT org>" <nospam@HOWLES.COM>
Subject:      Re: Eliminating rows, based on date values
Comments: To: tomas.corveleyn@GMAIL.COM

You've had other answers which will work, but here's my code (terse, tidy,
tested):

   data want;
   set have;
   by var_a;
   if dif(sasvalue)>=90 or first.var_a;
   run;

On Tue, 6 Dec 2005 07:29:28 -0800, Tomas <tomas.corveleyn@GMAIL.COM> wrote:

>Dear,
>
>I'm trying desperately to eliminate rows from the following type of
>file:
>
>obs       var_a Date    Sasvalue(date)
>  1             1       2/apr/02            15432
>  2             1       12/jun/03           15868
>  3             1       13/jun/03           15869
>  4             1       14/jun/03           15870
>  5             1       15/jun/03           15871
>  6             1       16/jun/03           15872
>  7             2       26/mar/02           15425
>  8             2       11/sep/03           15959
>  9             3       1/jan/03            15706
>
>With var_a being the id for a person. Based on the variable "Date" or
>the Sasvalue of that variable, I want to eliminate observations 3 to 6.
>I do not want to end up with only one observation per person, I want to
>keep the observations between which there are at least 90 days! So, my
>end-file has to be the following:
>
>obs       var_a Date    Sasvalue(date)
>  1             1       2/apr/02            15432
>  2             1       12/jun/03           15868
>  7             2       26/mar/02           15425
>  8             2       11/sep/03           15959
>  9             3       1/jan/03            15706
>
>I've tried "retain" statements working on the Sasvalue of the
>datevariable, but I haven't been able to compute the correct formula
>which eliminates only observations 3 to 6. Or I eliminate to much, or I
>eliminate to little. Can someone help me with this problem?
>
>Many thanks,
>
>Tomas Corveleyn.