My first Publication Agile-Data-Warehouse-Design-eBook | Page 81

60
Chapter 3
Dimensions
Dimensional
attributes are the
nouns and
adjectives that
describe events in
familiar business
terms
Dimensions are the significant nouns of a business or organization that form the
subjects, objects, and supporting details of interesting business events. They are 6
of the 7Ws: the who, what, when, where, why, and how of every event story. Dimen-
sional attributes further describe business events using terms that are familiar to
the stakeholders. They represent the adjectives that make data stories more inter-
esting. From a BI perspective, dimensions are the user interface to the data ware-
house, the way into the data. Dimensional attributes provide all the interesting
ways of rolling up and filtering the measures of business process performance. BI
applications use dimensions to provide the row headers that group figures on
reports and the lists of values used to filter reports. BI takes advantage of the
hierarchical relationships between dimensional attributes to support drill-down
analysis and efficient aggregation of atomic-detail measurements. The more
descriptive dimensional attributes you can provide, the more powerful the data
warehouse and BI applications appear. Consequently, good, richly descriptive
customer and product dimensions can have 50+ attributes.
The data values of a dimension (or an individual dimensional attribute) are re-
ferred to as its members.
Dimension Stories
Dimensions data
stories have weak
narratives. They are
subject and object
heavy but verb light
BEAM✲ modelers
add drama and plot
to help define
dimensions
Dimensions, because they represent descriptive reference data (adjectives and
nouns), lack the strong narrative of business events. Events (and event stories) are
associated with “exciting”, active verbs such as “buy”, “sell”, and “drive” as used in:
“customer buys product”, “employee sells service” and “James Bond drives Aston
Martin DB5”. Dimensions, on the other hand, are associated with static verbs such
as “has” and “is” that lead to weak narratives like “Customer has gender. Customer
has nationality” and “Product has product type. Product has storage”. These are
state of being events, archetypes for many “is/has” data stories such as “Vespa Lynd
is female. She is British” and “iPOM Pro is a computer. It has a 500GB disk”.
Important as these sentences are, we hardly think of them as stories because they
lack the drama of “who does what, to whom or what, and when, and where” that
propels you through all of the 7Ws to rapidly discover data requirements. Dimen-
sion stories are not exactly page-turners!
While data stories are highly effective at discovering dimensions and facts (as event
details) and the 7Ws remain a powerful checklist at all times, additional techniques
are needed to uncover the information hidden within the weaker narratives of
dimension stories. BEAM ✲ modelers have to add some drama to dimensions to
help stakeholders tell more interesting stories that fully describe their attributes
and business rules that affect ETL processing and BI reporting. To do this BEAM ✲
modelers use two plot devices: