[mailto:midrange-l-bounces@xxxxxxxxxxxx] On Behalf Of Dennis
Sent: Friday, June 03, 2011 12:40 PM
Subject: Identifying unique sets of combinations
We have a table as part of our HR system that is used to categorize days
off. For example, "S" (for sick) is in the SCKAC, USEDS, VACCK and 14 other
categories (I know, I know). V(acation) is in 7 categories, and so on.
These things are categorized in such a way because they intersect so much.
For example, a Carry-over vacation day and a Vacation day each have the same
effect in terms of Vacation Balance, but very different effects when it
comes to calculating next year's new balance.
In all, there are 110 absence codes that are grouped into 55 different
categories. In an effort to simplify all of this, I've been asked to
identify categories that share common members, with no exceptions. (In
other words, if categories AAA, ABB and ACC all contain V and C, and none of
them contain anything else, they are exact duplicates.)
By the same token, If V and G are in 5 categories, but a sixth category
lacks G, the powers that be would like this noted also.
I'm sure there's a way to get to this answer and still maintain what's left
of my sanity (many of you realize that there's little left), but I have not
yet landed on it. I thought SQL might get me to the answer most readily,
but try as I might I haven't found a workable solution yet; my CTEs quickly
get too complex for me to follow. (See prior note on sanity.) Pivot table
doesn't seem quite right (though it may help); counts are not definitive
So I thought I'd come to this group; see if you've had a similar situation,
and might be willing to share how you arrived at a solution.