logo for information-management-architect.com
Home
Strategy
Framework
Business Case Analysis
Project Planning
Requirements Analysis
Architecture & Design
Build Phase
Quality Assurance
Transition to Production
Management Information
Business Intelligence
Data Warehouse
Tools
Jobs
leftimage for information-management-architect.com

Dimension Data

Need dimension data for an information management requirements specification and want practical timesaving suggestions?

A dimensional model shows facts and dimensions, which are used for reporting and analytical purposes.  There is no logical representation of this model. It is a “physical” model but sometimes it is shown in a logical version in the modeling tool with English like names as opposed to physical names. 

It is sometimes called a “star schema” which permits multi-dimensional analysis.

Why is it different from a logical data model?

A logical model shows the business rules that apply to an organizations data. It contains entities, which are things of interest to a company and relationships, which are relationships between entities. Customer, order, product and sales territory might be considered entities and each entity may be described by many attributes or other entities.

A dimensional model contains facts and dimensions. The order is a fact and the customer, sales territory and date are dimensions. This means we can analyze orders by each of the dimensions that are related to order (In this case, order date, product and sales territory, and possibly sales representative).

A logical model depicts data in a “normalized” format. Essentially, this means that each data element only appears once on the model. (There are a many other technical ways to describe model normalization but that is outside the scope of requirements analysis).

A star schema allows duplication of data in the dimensions because this helps improve reporting/analysis performance. The following model is a simplified design for orders.

Dimension Data


In the example above, the order amount is the fact and this model lets us answer questions such as:
  • How many products were sold this week, this month, this quarter, etc;
  • How many products were sold to each customer group, etc;
  • How many products were sold by sales representatives, by sales territory, by sales area, etc
What is important for requirements analysis?

The business intelligence analyst needs to determine:
  • Which facts are required for analysis;
  • What questions need to be answered e.g. what query profiles are required to satisfy the business needs; and
  • How changing dimension data will be handled, e.g. suppose sales representative “A” makes a sale in April. Now suppose sales representative  "A" is replaced in June. Does the business want to keep track of the sales by each of the two sales representatives, or are they only concerned with sales by sales territory?
If they need to know sales by each representative, then we need to account for something called a “slowly changing dimension" and this needs to be included in the requirement specification.

Do we need a model for the requirements specification?

Not really! We need to know a list of facts and dimensions and a definition for each. However, a star schema design is a “physical” model, which is usually created in the architecture and analysis phase.

Summary…

Dimension data requirements are documented on a dimensional model, which shows analytical requirements for facts and dimensions. The logical data model is used to help derive dimensions.  It is important to understand analytic requirements, or at least the types of analytical questions, to complete the requirements specifications


footer for Information management page