On July 19th I gave a presentation to the Baltimore SQL Server Users Group titled "TSQL code to determine possible unique keys of raw data." The content can be found here. Here is the abstract:
I will be describing an algorithm I wrote that employs a hybrid recursive/iterative approach to determining the minimal unique key (if one exists), for a given table of raw data. The presentation will cover the problems involved in developing a solution and various approaches I considered, and of course a review of the solution at which I ultimately arrived.