Date: Wed, 23 May 2001 11:36:44 -0400
Reply-To: Charles Patridge <Charles_S_Patridge@PRODIGY.NET>
Sender: "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From: Charles Patridge <Charles_S_Patridge@PRODIGY.NET>
Subject: Re: PROGRAMMING QUESTION--Matching 2 datasets
I have a white paper (Fuzzy Merge/Match) on this subject which you can read
and download from the SAS Institute's Web site -
Included in this paper is actual SAS code that I used for the example within
the white paper. It will take a while to understand the actual SAS code and
modify it to meet your specific needs but the code can be modified to handle
your situation. Assuming Birth Month and Year and Gender to be additional
matching criteria will help extensively in creating "good" hits as well as
speeding up the process.
Not knowing your data I wonder how you would handle situations such as :
Elizabeth Patridge vs Liz Patridge vi Betty Patridge
as you can see the first name can cause no hits because of the use of
nicknames which is quite common on patient/customer records.
Hope this helps.
Some additional hyperlinks that might be useful:
Which will describe more about Fuzzy Merge, Data Cleansing, and other
commercial vendors who specialize in this arena.