Database Reference
In-Depth Information
For dimensional, we need to take the business questions we identified in the prior step and
then create a grain matrix. A grain matrix is a spreadsheet where the measures from the
business questions become columns and the dimensional levels from each business ques-
tion become rows.
The purpose of a grain matrix is to efficiently scope analytic applications. It is possible to
elicit hundreds of business questions, and after plotting these questions on a grain matrix,
we make the observation that questions from different departments can actually be very
similar to each other. By consolidating questions, we can scope applications that address
the needs for more than one department.
So each business question is parsed in terms of the measure or measures it contains, and the
levels of detail required for each measure. For example, recall the four university questions
we identified in Step 2:
1.
Show me the number of students receiving financial aid by department and
semester for the last five years. [From Financial Aid Office]
2.
Show me the number of students on full or partial scholarship by department and
semester for the last four years. [From Accounting Department]
3.
How many students graduated by department and semester over the last three
years? [From Alumni Affairs]
4.
How many students applied to the university over the last ten years? I want to com-
pare applications from high school students vs other universities. [From Admis-
sions Department]
Each of these questions comes from a different department yet focuses on the same meas-
ure, called
Student Count
. We would have to confirm that each area defines student the
same way; if yes, we can complete a grain matrix such as the following:
Student Count
Financial Aid Indicator
1
Semester
1, 2, 3
