Date:         Thu, 27 Nov 2008 21:45:30 -0800
Reply-To:     Alejandro <acajigal@GMAIL.COM>
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         Alejandro <acajigal@GMAIL.COM>
Organization: http://groups.google.com
Subject:      Re: delete multiple instances of same value in different fields?
Comments: To: sas-l@uga.edu
Content-Type: text/plain; charset=ISO-8859-1

On Nov 26, 7:05 pm, art...@NETSCAPE.NET (Arthur Tabachneck) wrote:
> Alejandro,
>
> I'm not sure if I correctly understand what you want to do.  Does the
> following come close?
>
> data have;
>   infile cards missover;
>   input
>        zip1  observed expected zip3
>        area1 area2 area3 area4 area5;
>   cards;
> 10001 10       12       10001 10002 10014 10023 10024
> 10002  0       0        10002 10001 10004 10029 10031 10055
> 10003  5       9        10003 10004 10005 10006 10012 10018
> 10001  0       0        10004 10003 10005 10007 10023
> ;
>
> proc sql noprint;
>   select distinct zip3 into :misses
>     separated by ' '
>       from have
>         where expected eq 0;
> quit;
>
> data want;
>   set have;
>   array area(5) area1-area5;
>   do i=1 to 5;
>     if area(i) in (&misses.) or
>      missing(area(i)) then
>      area(i)=99999;
>   end;
>   call sortn(of area[*]);
>   do i=1 to 5;
>     if area(i) eq 99999 then
>      call missing(area(i));
>   end;
> run;
>
> Art
> -------
>
> On Wed, 26 Nov 2008 15:05:49 -0800, Alejandro <acaji...@GMAIL.COM> wrote:
> >if I have a dataset where my expected = 0 , then I cannot use that zip
> >code.
> >i then need to remove that zipcode from every instance in my table and
> >replace it with a blank.
> >how do i do this?
>
> >this is for an analysis using a cluster detection scan statistic
> >called FlexScan.  FlexScan does not like missing expected values (no,
> >I can't just use 0.00001) and I need to delete every zip code in the
> >matrix file that it uses. The matrix file is the list of connected
> >zipcodes in the zip3, area1, area2, ..., area18 columns.
>
> >Here's an example (this is a made-up dataset for simplicity...these
> >zips aren't all connected in reality):
> >zip1     observed      expected     zip3     area1     area2
> >area3     area4   area5
> >10001   10               12             10001   10002    10014
> >10023     10024
> >10002    0                0               10002   10001   10004
> >10029     10031  10055
> >10003    5                9             10003  10004    10005    10006
> >10012  10018
> >10001    0                0             10004  10003    10005
> >10007     10023
>
> >In these 4 records, zips 10002 and 10004 both having missing
> >expecteds, so I need to delete zips 10002 and 10004 from every column.

wow, thanks Art.  that looks right.  i'll give it a try tomorrow and
let you know.