craig.bobchin (8/27/2015)
ScottPletcher (8/27/2015)
To me it seems fairly easy to match single rows. Maybe you can just output those matches to a table and then consolidate the like ids using that result.
SELECT m1.dimMemberId, m2.dimMemberId
FROM dimMember m1
INNER JOIN dimMember m2 ON
m1.dimMemberId < m2.dimMemberId AND
CASE WHEN LEFT(m1.FirstName, 3) = LEFT(m2.FirstName, 3) THEN 1 ELSE 0 END +
CASE WHEN m1.LastName = m2.LastName THEN 1 ELSE 0 END +
CASE WHEN m1.DOB = m2.DOB THEN 1 ELSE 0 END +
CASE WHEN m1.CurrentAddress1 = m2.CurrentAddress1 THEN 1 ELSE 0 END +
CASE WHEN m1.MemberCode = m2.MemberCode THEN 1 ELSE 0 END +
CASE WHEN m1.SSN = m2.SSN THEN 1 ELSE 0 END +
CASE WHEN m1.SubscriberCode = m2.SubscriberCode THEN 1 ELSE 0 END
>= 4
This is pretty close to what I'm looking for. I added your join login in the select as a new column called Score. Now I just need to figure out how to assign a common value to a new field based on the matches.
Scott did it pretty much as I described including being able to eliminate roughly half the work by using a Triangular Join instead of a Full (Square) Cartesian Join.
The question now is, what do you mean by a "common value"?
--Jeff Moden
Change is inevitable... Change for the better is not.