Question:

How do you use the Sequence Clustering Algorithm in SQL Server 2005 (Yukon) Analysis Services?

Answer:

Sequence clustering is a hybrid algorithm between clustering and sequence. In order to use Sequence Clustering, you need to have two tables: (1.) A case table containing at least a case id (2.) A nested table containing at least three columns: a. case id(foreign key, do not include in the mining structure) b. sequence id (key)  c. sequence state (input/predict). In SQL Server 2005 Beta 2 CD, there is an AS sample project,which includes a model for sequence clustering on Adventure works.

Resource(s):

http://www.sqlserverdatamining.com

Sign-up For Data Mining Source Code Newsletter