Moss L.T., Atre S. Business intelligence roadmap: The complete project lifecycle for decision-support applications
Подождите немного. Документ загружается.
[ Team LiB ]
Bibliography and Additional Reading
Adelman, Sid, and Larissa Terpeluk Moss. Data Warehouse Project Management.
Boston, MA: Addison-Wesley, 2000.
Cockburn, Alistair. Writing Effective Use Cases. Boston, MA: Addison-Wesley, 2000.
Dyché, Jill. e-Data: Turning Data into Information with Data Warehousing. Boston,
MA: Addison-Wesley, 2000.
English, Larry P. Improving Data Warehouse and Business Information Quality:
Methods for Reducing Costs and Increasing Profits. New York: John Wiley & Sons,
1999.
Hoberman, Steve. Data Modeler's Workbench: Tools and Techniques for Analysis and
Design. New York: John Wiley & Sons, 2001.
Imhoff, Claudia, Lisa Loftis, and Jonathan G. Geiger. Building the Customer-Centric
Enterprise: Data Warehousing Techniques for Supporting Customer Relationship
Management. New York: John Wiley & Sons, 2001.
Inmon, William H., Claudia Imhoff, and Ryan Sousa. Corporate Information Factory.
New York: John Wiley & Sons, 1997.
Jackson, Michael. Software Requirements and Specifications: A Lexicon of Practice,
Principles, and Prejudices. Reading, MA: Addison-Wesley, 1995.
Jarke, Matthias, Maurizio Lenzerini, Yannis Vassiliou, and Panos Vassiliadis.
Fundamentals of Data Warehouses. New York: Springer, 2000.
Kimball, Ralph, Laura Reeves, Margy Ross, and Warren Thornthwaite. The Data
Warehouse Lifecycle Toolkit: Expert Methods for Designing, Developing, and
Deploying Data Warehouses. New York: John Wiley & Sons, 1998.
Kovitz, Benjamin L. Practical Software Requirements: A Manual of Content and Style.
Greenwich, CT: Manning Publications Company, 1998.
Moeller, R. A. Distributed Data Warehousing Using Web Technology: How to Build a
More Cost-Effective and Flexible Warehouse. New York: AMACOM American
Management Association, 2001.
Ross, Ronald G. The Business Rule Concepts. Houston, TX: Business Rule Solutions,
Inc., 1998.
Von Halle, Barbara. Business Rules Applied: Building Better Systems Using the
Business Rules Approach. New York: John Wiley & Sons, 2001.
Wiegers, Karl E. Software Requirements. Redmond, WA: Microsoft Press, 1999.
Wood, Jane, and Denise Silver. Joint Application Development. New York: John Wiley
& Sons, 1995.
北斗成功社区 BeiDouWeb.com 教育音视频/电子书/实用资料文档/励志音乐影视 仅供免费试用/版权原著所有
171/631
Yourdon, Edward. Death March. Upper Saddle River, NJ: Prentice Hall, 1997.
[ Team LiB ]
北斗成功社区 BeiDouWeb.com 教育音视频/电子书/实用资料文档/励志音乐影视 仅供免费试用/版权原著所有
172/631
[ Team LiB ]
Chapter Five. Step 5: Data Analysis
北斗成功社区 BeiDouWeb.com 教育音视频/电子书/实用资料文档/励志音乐影视 仅供免费试用/版权原著所有
173/631
北斗成功社区 BeiDouWeb.com 教育音视频/电子书/实用资料文档/励志音乐影视 仅供免费试用/版权原著所有
174/631
Chapter Overview
This chapter covers the following topics:
Things to consider when analyzing source data for BI applications
The difference between the systems analysis phase of a traditional methodology
and the business-focused data analysis performed during this step
Top-down logical data modeling, including project-specific logical data models,
integrated enterprise logical data models, and data-specific business meta data
components gathered during the logical data modeling process
Bottom-up source data analysis, including how to apply three sets of
transformation rules to the source data: technical data conversion rules,
business data domain rules, and business data integrity rules
The responsibilities for data archeology, data cleansing, and data quality
enforcement, plus the need to triage (prioritize) data-cleansing activities
Brief descriptions of the activities involved in data analysis, the deliverables
resulting from those activities, and the roles involved
The risks of not performing Step 5
Things to Consider
Source Data
Do we know where the source data resides? In what systems? In
what files? In what databases?
Are there multiple potential sources for the same data?
Has the requested source data already been modeled?
How current is the business meta data on those models?
Have the data owners ratified the business meta data?
Do we know who the data owners are? Who has authority over the
source data?
Is there any other type of documentation available for the
requested source data? Is it current and complete?
Where is that documentation? In a meta data repository? In
programs? In manuals?
Data Quality
Do we know how clean the source data is?
北斗成功社区 BeiDouWeb.com 教育音视频/电子书/实用资料文档/励志音乐影视 仅供免费试用/版权原著所有
175/631
How clean does the data have to be according to our business
representative?
Will that be clean enough for other knowledge workers, business
analysts, and business managers who will use the same data?
Do we know who they are?
Where do we get the business rules for the data? From the data
owners? From the business representative on the project?
Data Cleansing
Have data errors already been documented by other project teams?
If so, where is that documentation?
Who would know what the known data errors are?
Are codes being translated inside operational programs? If so, in
which programs?
Does a code translation book exist for encoded fields?
Do we already know which data is critical, which is important, and
which is insignificant (for data-cleansing triage purposes)?
Operational systems are developed as stovepipe automation solutions for individual
business units and not as support for the executive decision-making process.
Therefore, operational systems are not designed to integrate or reconcile with each
other in order to provide a consistent cross-organizational view. BI applications, on
the other hand, are designed to do just that—provide integrated and reconciled
business data to the business community.
[ Team LiB ]
北斗成功社区 BeiDouWeb.com 教育音视频/电子书/实用资料文档/励志音乐影视 仅供免费试用/版权原著所有
176/631
[ Team LiB ]
Business-Focused Data Analysis
For many organizations, the BI decision-support initiative is the first attempt to bring
business data together from multiple sources in order to make it available across
different departments. Organizations that use a traditional systems development
methodology on their BI projects usually run into severe source data problems when
they try to implement their extract/transform/load (ETL) processes because
traditional development methodologies do not have steps for analyzing data domains
early in the development process. They have, at best, a systems analysis phase for
the application functions but no business-focused data analysis phase for the
underlying data.
The business-focused data analysis step is the most
critical cross-organizational step described in Business
Intelligence Roadmap.
Step 5, Data Analysis, is different from a systems analysis phase in a traditional
methodology. The activities traditionally performed during systems analysis are
geared toward producing a design decision for the system to be built. The activities
performed during data analysis are geared toward understanding and correcting the
existing discrepancies in the business data, irrespective of any system design or
implementation method. Data analysis is therefore a business-focused activity, not a
system-focused activity.
Figure 5.1 points out that two complementary methods are required to perform
rigorous data analysis:
Top-down logical data modeling for integration and consistency1.
Bottom-up source data analysis for standardization and quality2.
Figure 5.1. Complementary Data Analysis Techniques
北斗成功社区 BeiDouWeb.com 教育音视频/电子书/实用资料文档/励志音乐影视 仅供免费试用/版权原著所有
177/631
[ Team LiB ]
北斗成功社区 BeiDouWeb.com 教育音视频/电子书/实用资料文档/励志音乐影视 仅供免费试用/版权原著所有
178/631
[ Team LiB ]
Top-Down Logical Data Modeling
The most effective technique for discovering and documenting the single cross-
organizationally integrated and reconciled view of business data is entity-relationship
(E-R) modeling, also known as logical data modeling. A popular approach to E-R
modeling in the early 1980s was to model all the data for the entire organization all
at once. While this approach was a worthwhile architectural endeavor, it did not yield
better systems because the process was not integrated with the systems
development lifecycle. A more effective approach is to incorporate E-R modeling into
every project and then merge the project-specific logical data models into one
consolidated enterprise data model over time.
Project-Specific Logical Data Model
E-R modeling is based on normalization rules, which are applied during top-down
data modeling as well as during bottom-up source data analysis. Using normalization
rules along with other data administration principles assures that each data element
within the scope of the BI project is uniquely identified, correctly named, and
properly defined and that its domain is validated for all business people who will be
accessing the data. Thus, the normalized project-specific logical data model yields a
formal representation of the data exactly as it exists in the real world, without
redundancy and without ambiguity.
This formal representation of data follows another normalization rule: process
independence. Therefore, by definition, a logical data model, which is based on
normalization rules, is also process independent. Process independence means that
the structure and content of the logical data model are not influenced by any type of
database, access path, design, program, tool, or hardware, as shown by the X
markings in Figure 5.2.
Figure 5.2. Process Independence of Logical Data Models
北斗成功社区 BeiDouWeb.com 教育音视频/电子书/实用资料文档/励志音乐影视 仅供免费试用/版权原著所有
179/631
Because of its process independence, a logical data model is a business view, not a
database view and not an application view. Therefore, a unique piece of data, which
exists only once in the real business world, also exists only once in a logical data
model even though it may be physically stored in multiple source files or multiple BI
target databases.
Enterprise Logical Data Model
It is the responsibility of an enterprise architecture group, or of data administration if
the organization does not have an enterprise architecture group, to merge the
project-specific logical data models into an integrated and standardized enterprise
logical data model, as illustrated in Figure 5.3.
Figure 5.3. Creating an Enterprise Logical Data Model
北斗成功社区 BeiDouWeb.com 教育音视频/电子书/实用资料文档/励志音乐影视 仅供免费试用/版权原著所有
180/631