End User Applications

1
End User Applications

• Lecture 1

2
Methodology Per Development Increment
Business Needs
Delivery
Storage
Engineering
Acquisition
Access

• Define business requirements, with focus on
  information requirements
• Define dimensional data model for the data mart
  facts and dimensions
• Define data model for the DW
• Identify authoritative source of required data
  from operational systems and analyze source data
• Map source data, define transformation
  requirements
• Design build and test extract, transformation and
  load mechanisms
• Design and build end user application
• Demonstrate and refine user application via
  prototyping techniques
• Train and conduct user acceptance

3
Data Access
Acquisition
Engineering
Storage
Data Delivery
Access
Metadata
Purpose The end user environment that
'consumes' data for reporting, analysis and
knowledge discovery
4
Classification of end user applications

• We can generally classify end user applications
  as follows
• Business user-oriented Users generally work in
  some area of an organization and use a data mart
  and data access tool to support the work they do.
• Statistically oriented the data warehouse
  environment is used by a small, highly-skilled
  group of users to develop statistical models.
  These models are used to support various business
  activities (like sales and marketing).
• The process we use to meet the requirements of
  these users tends to be quite different

5
Business-user Oriented environments

• Architecture
• Dimensional data model
• RDBMS or MDDB
• Data mart is pre-built and refreshed periodically
  
• Classes of tools
• OLAP
• ROLAP
• Managed Query
• Report Writers

6
Business-user Oriented environments

• Wide range of users supported
• Power users/knowledge workers
• Ad hoc capabilities
• Often work within a department to create reports
  and report templates for other, less
  sophisticated users
• Few limitations (other than restricting access to
  underlying data based on need)
• Managers
• Use data mart to make day to day decisions
• Want easy to use, but flexible interface
• Standard business users
• Want pre-built reports to support day to day
  activities

7
Business-user Oriented Development

• For each group of users we need to define, design
  and implement application templates
• Application templates define the structure and
  layout of on-line and hard copy reports and
  queries
• The degree detail of structure and layout will
  vary by user group little for power users (they
  define their own), lots for standard users.
• Most common template parameterized reports

8
Business-user Oriented Development

• Assume data mart has been designed
• Step 1 Set up the environment
• Choose, test and implement access tool(s)
• Define design documentation standards
• Define naming conventions for queries,reports,
  report templates, etc.
• Define methods for ensuring design documentation
  is placed in metadata repository

9
Business-user Oriented Development

• Step 2 Determine the Initial Template Set for
  each user group
• Identify candidate reports based on user
  requirements
• Classify the reports by use within analysis
  cycle
• Hows business
• Trends (time series)
• Whats unusual (exception reporting)
• Exception drivers
• What if
• Decision making
• Prioritize the list of reports identify the
  starter set

10
Business-user Oriented Development

• Step 3 Design Navigation Approach
• Need to determine how reports will be classified
  and organized for access indexing strategy
• Organization becomes an issue when the number of
  reports grows large need to think about it up
  front so change not required later
• Step 4 Determine Template Standards
• Naming of labels
• Location of objects (headers, footers, etc.)
• Formatting of reports
• Look and feel

11
Business-user Oriented Development

• Step 5 Develop Detailed Template Specifications
• Design each of the templates

12
Business-user Oriented Development

• Step 6 Select Implementation approach
• Paper
• Web
• Direct access to tool (Brio, Business Objects)
• Custom-coded interface
• Step 7 Develop prototype, refine
• Step 8 Develop production version, implement

13
Statistical-user Oriented Design

• Statistical users have very different
  requirements
• Typically, these users need to explore all the
  data in the warehouse, looking to develop
  different types of statistical models for
• Describing customers
• Predicting behavior (marketing)
• Identifying fraud

14
Statistical users
Data Warehouse
1

• Tools
• Managed Query
• Statistical Analysis
• Data Mining

Data Mart
2

• Typical Process
• Standard load to analytic data mart of commonly
  used data (typically aggregated data)
• Users given the ability to request data directly
  from the data warehouse, as required to perform
  statistical analysis. Typically, analysts need
  atomic level data.

15
Statistical Users

• Design goals
• Provide required aggregated data
• Provide access to core data warehouse ensure no
  conflict with ETL and data distribution process
• Provide proper tools