Physical Database Design

320 CHAPTER 14 Physical Design for Decision Support, Warehousing, and OLAP

respectively. Generally, aggregating data from one view into another view increases the

data density.

The difference in data density from the core data to summary views can be very

marked. Extending our grocery store example, let’s say that the store has 3,000 items

available, and the average customer buys 30 distinct items per visit, shopping once a

week. The data density of the core data is (30/3,000)(1/5) = 0.002. However, if we sum-

marize the data across all customers, then it would not be unreasonable to have a data

density of say 0.95 (i.e., most items have some units ...