Date:         Mon, 10 Apr 2000 09:54:01 -0400
Reply-To:     Steve Giblin <Steve_Giblin@VAPOWER.COM>
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
Comments:     To: Victor Aina <aina@SFU.CA>
From:         Steve Giblin <Steve_Giblin@VAPOWER.COM>
Subject:      Re: Match merge problem
Comments: To: SAS-L@LISTSERV.VT.EDU
Content-type: text/plain; charset=us-ascii

Another approch you may wish to employ is SQL.
  This will allow you to grab multiples of partial matches,
   then you can devise a hierarchy for which matches are "best".

  Note:  You will have to be clear on your variable names
               You might also want to have a unique sequence number
                in each of the lists (ex SEQA, SEQB) to help in the dedupe
                process.


  Something like

   PROC SQL;
             CREATE TABLE MATCH AS
              SELECT A.*, B.*
                  FROM LISTA A, LISTB B
                WHERE A.PHONEA = B.PHONEB
                           OR A.LNAMEA = B.LNAMEB
                           OR A.ADDR1A    = B.ADDR1B

                            OR ...

     ; QUIT;

      * If you have the name components parsed in one file but not the other ;
      * you can employ something like

                            OR INDEX(FULLNAMA,TRIM(LNAMEB))>0

  * Same thing for address components

                            OR INDEX(ADDR1A,TRIM(HOUSENMB))=1
                            OR INDEX(ADDR1A,TRIM(STREETB))>0


   This SQL method will allow you to grab all of the  possible matches and then
you will
    have to read through multiples and settle upon what is "best".
    You will need to makes sure that you do not apply too loose of an OR
condition,
     otherwise you will get too many.   For example, if you simply allow for a
lookup of
     "Smith", on a last name match, you will get many unwanted matches.  But if
you
     do "Smith"  AND parts of the address, you can manage your result set.

     Hope this adds another option for you.

      Steve