Page 102 - DCAP606_BUSINESS_INTELLIGENCE

Unit 6: Business Intelligence Project

Notes
Obtain Existing Documentation
When seeking to understand a data source, the first thing to do is look at existing
documentation. This avoids “re-inventing the wheel”. If a data source is fully documented,
data profiled and of high quality most of the job of data source discovery is complete.

Existing documentation may include:
 Data models
 Data dictionary

 Internal/technical documentation
 Business user guides
 Data profiles and data quality assessments
Check through the documentation to assess its completeness and usefulness.
The data source analyst should study the existing documentation before any in depth
discussions with the SMEs. This improves the credibility of the data analyst and save time
for the SMEs.

Model and Define the Input
The data model is a graphic representation of data structures that improves understanding
and provides automation linking database design to physical implementation. This section
assumes that the data source is stored in a relational database that modelled using typical
relational data modelling tools.

If there is an existing data model, start with that, otherwise use the reverse engineering
capability of the data modelling to build a physical data model. Next, group the tables that
are of interest into a subject area for analysis. Unless, a large percentage of the data source
is needed for the data warehouse avoid studying the entire data source. Stay focused on
the current project.
For each selected data source table define:
 Physical Name

 Logical Name
 Definition
 Notes
For each selected data source column define:

 Physical Name
 Logical Name
 Order in Table
 Datatype

 Length
 Decimal Positions
 Nullable/Required
 Default Value
Contd....

LOVELY PROFESSIONAL UNIVERSITY 97