Data Mining with Microsoft® SQL Server® 2008

Chapter 13. Mining OLAP Cubes

You may have already heard of (or even have experience with) Online Analytical Processing (OLAP). E. F. Codd, the originator of the relational data model, wrote a white paper in 1994 that introduced the term Online Analytical Processing into the lexicon of database users. OLAP is the current term for systems that were formerly called decision-support systems (DSS) or multidimensional databases.

OLAP plays an important role in today's business intelligence (BI) market. An OLAP database contains a number of cubes, similar to the way a relational database contains a number of tables. As you will see in this chapter, a cube has a set of well-defined dimensions and measures. A dimension is an analytical object that provides axes on which you can ask questions, whereas a measure is a numerical value that exists at a coordinate in a cube.

For example, almost all analytical cubes have a Time dimension, and a supermarket chain's sales cube may have additional dimensions, such as Store, Product, and Customer. Each dimension can also have one or more hierarchies. The Product dimension may have a category and subcategory hierarchy. Individual members in these dimensions identify a coordinate where you can examine measures, such as Unit Sales, Store Sales, Profit, and Cost. For example, you could ask the value of Profit in the Redmond store for Canned Vegetables during August of 2008—the dimensions Store, Product, and Time specify the coordinate for the measure Profit. ...