Transcript Chapter 8 Constructing a Decision Support System and DSS

CHAPTER 8
Decision Support Systems
Development
1
8.1 DSS Development Issues


How to develop a DSS
DSS must usually be custom tailored
– The application are diverse, ranging from dataoriented DSS to model-oriented DSS in different
functional areas
– The vendors assisted in the DSS Construction
– Hardware, networking, man-machine interface and
potential impact of DSS on the individual and
groups. Software problem focused in this chapter
– Other problem are:
Fall, 2006
All Right Reserved, Zhong YAO, School of E&M, BHU
8-2
8.1 DSS Development Issues
–
–
–
–
–
–
–
–
–
Fall, 2006
System development life cycle (SDLC)
Prototyping
Objective-Oriented Developing Method
ROMC
Organizing and forming the development team
Complex process
Technical issues
Behavioral issues
Different approaches
All Right Reserved, Zhong YAO, School of E&M, BHU
8-3
8.2 DSS Development Platform







General-purpose programming language, such as
COBOL or PASCAL. Little has been used in 1990s
Fourth-generation language (4GL), like a SQL.
OLAP with a data warehouse or large database
DSS integrated development tool (generator, engine),
such as Excel, Lotus Domino.
Domain-specific DSS generator, such SAS, MAPLE,
now Mathmatica, Matlab.
Use the CASE methodology
Integrate several of the above
Fall, 2006
All Right Reserved, Zhong YAO, School of E&M, BHU
8-4
8.3 Traditional Systems Development
Life Cycle (SDLC) (Waterfall)
Need
Planning
Analysis
Design
Implementation
System
Fall, 2006
All Right Reserved, Zhong YAO, School of E&M, BHU
8-5
8.3 Traditional Systems Development
Life Cycle (SDLC)
–
–
–
–
–
–
–
–

Phase A-Planning,
Phase B-Research,
Phase C-System Analysis and Conceptual Design,
Phase D-Design,
Phase E-Construction,
Phase F-Implementation,
Phase G-Maintenance and Documentation,
Phase H-Adaptation
Simplifying Into Four Phases:
–
–
–
–
Fall, 2006
Planning (Phase A, and Phase B)
Analysis (Phase C)
Design (Phase C , D, and E)
Implementation (Phase F, G, and H)
All Right Reserved, Zhong YAO, School of E&M, BHU
8-6
8.3 Traditional Systems Development
Life Cycle (SDLC)

Planning - Why Build the System?
Minor Step
Deliverable
1. Identify business value
2. Analyze feasibility
3. Develop work plan
4. Staff project
System request
Feasibility study
Work plan
Staffing plan,
Project charter
Project management tools
CASE tool
Standards list
Project binders / files
Risk assessment
5. Control and direct project
Fall, 2006
All Right Reserved, Zhong YAO, School of E&M, BHU
8-7
8.3 Traditional Systems Development
Life Cycle (SDLC)

Analysis- Who, What, When, Where?
Minor Step
Deliverable
6. Analyze problem
Analysis plan
7. Gather information
Information
8. Model process(es)
Process model
9. Model data
Data model
Fall, 2006
All Right Reserved, Zhong YAO, School of E&M, BHU
8-8
8.3 Traditional Systems Development
Life Cycle (SDLC)

Design - How Will the System Work?
Minor Step
Deliverable
10. Design physical system
Design plan
11. Design architecture
Architecture design,
Infrastructure design
12. Design interface
Interface design
13. Design database and files
Data storage design
14. Design program(s)
Program design
Fall, 2006
All Right Reserved, Zhong YAO, School of E&M, BHU
8-9
8.3 Traditional Systems Development
Life Cycle (SDLC)

Implementation--- System Delivery
Minor Step
Deliverable
15. Construction
Test plan,
Programs,
Documentation
16. Installation
Conversion plan,
Training plan
Fall, 2006
All Right Reserved, Zhong YAO, School of E&M, BHU
8-10
8.3 Traditional Systems Development
Life Cycle (SDLC)

Common Implementation Headaches
–
–
–
–
–
–
–
–
Fall, 2006
No project team or management support
Hazy purpose; no defined schedule; ballooning scope
Unclear aspects of make vs. buy decisions
Few project integrations are functional out of the box
Qualitative benefits
No user buy in
Poor project management skills
No accountability (责任) / no responsibility
All Right Reserved, Zhong YAO, School of E&M, BHU
8-11
8.3 Traditional Systems Development
Life Cycle (SDLC)

CASE Tools Functions
–
–
–
–
–

Information systems for systems analysts
Can help manage system development
Upper CASE (assists in analysis)
Lower CASE (manages diagrams and code generation)
Integrated CASE (both)
Often used tools:
–
–
–
–
–
–
–
–
Fall, 2006
Oracle Enterprise Development Suite
Rational Rose
Paradigm Plus
Visible Analyst
Logic Works Suite
AxiomSys and AxiomDsn
V32 & X32
Visual Studio
All Right Reserved, Zhong YAO, School of E&M, BHU
8-12
8.3 Traditional Systems Development
Life Cycle (SDLC)


Fall, 2006
Project Management (PM)
– Team leader must have good PM skills
– Major reason for IS development failures-bad PM skills
– Only 26% of all projects surveyed (23,000) in 1998
succeeded
– 28% failed, 46% challenged
– Lower success rates for large companies
– Better PM skills needed
Skills for Project Managers
– Technology and business knowledge
– Judgment
– Negotiation
All Right Reserved, Zhong YAO, School of E&M, BHU
8-13
8.3 Traditional Systems Development
Life Cycle (SDLC)

– Good communication
– Organization
Implementation Failures (DW Example)
– No user involvement
– No clear objectives stated early
– No real executive sponsorship
– Not appropriate for the DSS development
Fall, 2006
All Right Reserved, Zhong YAO, School of E&M, BHU
8-14
8.4 Alternative Development
Methodologies


Parallel development
Rapid application development (RAD) methodologies
– Phased development
– Prototyping
– Throwaway prototyping
Fall, 2006
All Right Reserved, Zhong YAO, School of E&M, BHU
8-15
8.4 Alternative Development
Methodologies


Parallel Development
– Multiple copies of design and implementation phases
– To develop separate subsystems
– All come together in a single implementation phase
Phased Development
– Break system up into versions developed sequentially
– Each version has more functionality
– Evolves into a final system
– Users gain functionality quickly
– But initial systems are incomplete
Fall, 2006
All Right Reserved, Zhong YAO, School of E&M, BHU
8-16
8.4 Alternative Development
Methodologies
Prototyping (also called Evolutionary Prototyping
Process, iterative process, middle-out process, adaptive
design, incremental design) Characteristics:
– Performing analysis, design, and implementation
phases concurrently, and repeatedly
– Users see system functionality quickly and provide
feedback
– Decision maker learns about problem
– But can lose gains in repetition
 Aims: building a DSS in a series of short steps with
immediately feedback from users to ensure that
development is proceeding correctly.

Fall, 2006
All Right Reserved, Zhong YAO, School of E&M, BHU
8-17
8.4 Alternative Development
Methodologies

Developing processes:
– Select an important sub-problem to be built first. User
and the builder jointly identify a subproblem for which
the initial DSS is constructed. This early joint effort sets
up initial working relationships between the
participants and opens the lines of communication. The
subproblem should be small enough that the nature of
the problem, the need for computer-based support, and
the nature of the at support are clear. It should have
high interest value to the decision maker even if that
interest is short-live.
Fall, 2006
All Right Reserved, Zhong YAO, School of E&M, BHU
8-18
8.4 Alternative Development
Methodologies
– Develop a small but usable system for the decision
maker. No major system analysis or feasibility analysis
is involved. In fact, the builder and the user go through
all the steps of the system development process quickly,
though on a small scale. The system should, out of
necessarily, be simple.
– Evaluate the system constantly. At the end of each cycle
the system is evaluated by the user and builder.
Evaluation is an integral part of the development
process, and is the control mechanism for the entire
iterative design process. The evaluation mechanism is
what keeps the cost and effort of developing a DSS
Fall, 2006
All Right Reserved, Zhong YAO, School of E&M, BHU
8-19
8.4 Alternative Development
Methodologies
– consistent with its value. At the end of the evolution a
decision is made on whether to further refine the DSS,
or to stop.
– Refine, expand, and modify the system in cycles.
Subsequent cycles expand and improve the original
version of the DSS. All the analysis, design,
construction, implementation, and evaluation steps are
repeated in each successive refinement.

Advantages of Prototyping
– Short developing time
– Short user reaction time
– Improve user understanding of the system, its
information needs, and its capabilities.
Fall, 2006
All Right Reserved, Zhong YAO, School of E&M, BHU
8-20
8.4 Alternative Development
Methodologies
– Low cost.

Disadvantages and Limitations
– Gains maybe lost.
• Gains includes: Understanding Information systems
benefits and costs, a detailed description of the
business’s information needs, an easy to to maintain
information system design, a well-tested information
system, and well-prepared users.
– Combined with the critical success factor method
• Depend on the DSS built by the end-user or a DSS
team.
Fall, 2006
All Right Reserved, Zhong YAO, School of E&M, BHU
8-21
Prototyping
Need
Planning
Analysis
Design
Implementation
Prototype
Prototype Not OK
Prototype OK
System
Fall, 2006
All Right Reserved, Zhong YAO, School of E&M, BHU
8-22
8.4 Alternative Development
Methodologies


Throwaway Prototyping
– Like prototyping and SDLC
– Analysis phase is thorough
– Design prototypes assist in understanding the system
– Example: can use Excel, then Visual Basic
Prototyping for DSS Development
– Problems are semistructured or unstructured
– Managers and developers may not completely understand
problem
– Use prototyping
Fall, 2006
All Right Reserved, Zhong YAO, School of E&M, BHU
8-23
Throwaway Prototyping (Figure 6.4)
Need
Planning
Analysis
Design
Design
Design Prototype
Not OK
Implementation
Implementation
System
Design
Prototype
Fall, 2006
All Right Reserved, Zhong YAO, School of E&M, BHU
8-24
8.4 Alternative Development
Methodologies

Why Prototyping?
– Users and managers involved in every phase and
iteration
– Learning is part of design
– Prototyping bypasses the information requirement
definition
– Short interval between iterations
– Initial prototype must be low cost
Fall, 2006
All Right Reserved, Zhong YAO, School of E&M, BHU
8-25
8.5 Team-Developed DSS

A team-Developed DSS needs:
–
–
–
–
Fall, 2006
Substantial effort.
Extensive planning and organization
Some generic activities
Group of people to build and to manage it
(users, intermediaries, DSS builder, technical support
experts and IS personnel)
Size depends on
• Effort
• Tools
• For example, some project needs 2-3 people, but
other maybe 15-20 people.
All Right Reserved, Zhong YAO, School of E&M, BHU
8-26
8.5 Team-Developed DSS


DSS development group varies:
– Within the IS department
– As a highly placed executive staff group
– Within the finance or other functional area
– Within the industrial engineering department
– Within the management science group
– Within the information center group
The process that a DSS team may follow depends on the
specific application. The group may be temporary, created
for a specific DSS, or it may be permanent, in which case
the group members are assigned to specific DSS project.
Fall, 2006
All Right Reserved, Zhong YAO, School of E&M, BHU
8-27
8.6 Team-Developed Versus UserDeveloped DSS


DSS 1970s and early 1980s were large-scale, complex systems
designed primarily to provide organizational support. Therefore,
these process needs team effort to complete and maintain.
Another approach is user-developed system.
–
–
–
–
–
–
–
–
–
–
–
Fall, 2006
Personal computers
Computer communication networks
PC-mainframe communication
Friendly development software
Reduced cost of software and hardware
Increased capabilities of personal computers
Enterprise-wide computing
Easy accessibility to data and models
Client/server architecture
Now OLAP
Balance
All Right Reserved, Zhong YAO, School of E&M, BHU
8-28
8.7. End-user Computing and
User-Developed DSS

End-user Computing (end-user development):
development and use of computer-based information systems
by people outside the formal information systems areas.
This definition includes many people, such as manager,
professionals using PCs, Secretaries using Word processing
tools, etc.

End-users Can be
–
–
–
–
Fall, 2006
At any level of the organization
In any functional area
Levels of computer skill vary
Growing
All Right Reserved, Zhong YAO, School of E&M, BHU
8-29
8.7. End-user Computing and
User-Developed DSS


End-User Developed DSS Advantages
–
Short delivery time
–
Eliminate extensive and formal user requirements specifications
–
Reduce some DSS implementation problems
–
Low cost
End-User Developed DSS Risks
–
Poor Quality
–
Quality Risks
 Substandard or inappropriate tools and facilities
 Development process risks
 Data management risks
–
Increased Security Risks
–
Problems from Lack of Documentation and Maintenance
Procedures
Fall, 2006
All Right Reserved, Zhong YAO, School of E&M, BHU
8-30
8.7. End-user Computing and
User-Developed DSS

Issues in Reducing End-User Computing Risks
– Error detection
– Use of auditing techniques
– Determine the proper amount of controls
– Investigate the reasons for the errors
– Solutions
– Spreadsheet errors
• Should use same controls as normal IS
Fall, 2006
All Right Reserved, Zhong YAO, School of E&M, BHU
8-31
Developer Attributes
•Age
•Gender
•Computer Confidence
•Domain Experience
•Application Expertise
•Development Expertise
•Math anxiety
•Cognitive style
Application Type
•Spreadsheet
•Database
•Others
End-user development
Application Outcomes
•Reliability
•Ease of use
•Maintainability
•Auditability
•Cost
Problem/process Char.
•Problem Complexity
•Time Pressure
•Existence of Review
•Other
Developer Approach
•Ad Doc
•Structured
Fall, 2006
Developer Configuration
•Singles
•Pairs
•3+
All Right Reserved, Zhong YAO, School of E&M, BHU
8-32
8.8. Objective-Oriented Developing
Methods For DSS

Evolution of Objective-Oriented Technology
– Simula Language for replacing Simulation Language
– Use of auditing techniques (1967)
– Ada and Midula-2 for information hiding, (70s)
– Smalltalk, Xerox PARC, 1980
– C++, AT and T, 1981
– Visual Basic, Power Builder, Delphi etc. since 1994
– Java, C#, etc. 2000s
Fall, 2006
All Right Reserved, Zhong YAO, School of E&M, BHU
8-33
8.8. Objective-Oriented Developing
Methods For DSS

Main Concept of Objective-Oriented Technology
– To analyze and resolve the problems, the initial views
of the OOT is according to human being recognizing
objective world and thinking ways
– Objective world consists of many concrete things or
events, abstract concept, planning etc.
– Therefore, to study the problem of objective world, we
can abstract them as objects. In OOT, the objects are
basic elements. It is a core of analyzing problem.
– What is Object?
• Is of independent operations and behaviors program.
• Object vs. procedure.
Fall, 2006
All Right Reserved, Zhong YAO, School of E&M, BHU
8-34
8.8. Objective-Oriented Developing
Methods For DSS

What is Object?
– Actions executed by messages transferred between objects.
– Each object possesses itself data structure and style for storing
data
– Each object possesses independent interface for accessing by
other objects
Object 1
Object 2
Object 4
Fall, 2006
User demand
Object 3
Object 5
Object 6
All Right Reserved, Zhong YAO, School of E&M, BHU
8-35
8.8. Objective-Oriented Developing
Methods For DSS

Object Class
– If a object having same structure, operation, and the
following same constraint rules
– A class include: (1) name, (2) external interface, (3)
internal presentation and (4) realization.
– Characteristics:
– Class Specification---interface
– Realization--- each function how to do
– Hierarchy (Lattice)---farther class, sub-class, sub-sub-class etc.
» High level: generalization and commonality
» Low level: specials and details
Fall, 2006
All Right Reserved, Zhong YAO, School of E&M, BHU
8-36
8.8. Objective-Oriented Developing
Methods For DSS

Characteristics of Objective-Oriented Technology
– Abstraction: Any thing can be presented, structured and
unstructured. For example, table, chair, graph, sound, rule,
concept etc. But similar things will be abstracted into same
class for their similar operations, structures, and limits)
– Encapsulation: structure and procedure encapsulated into
one whole class. This can be used for data hiding. Data in a
class only can be accessed by its (this class) internal
functions or data.
– Inheritance: subclass can inherit partial or whole properties
of its mother class. The subclass can extend some special
properties that the mother class don’t have.
– Polymorphism: one function can be used by different
arguments.
Fall, 2006
All Right Reserved, Zhong YAO, School of E&M, BHU
8-37
8.9. UML Technology for DSS

Resources: Universal Modeling Language
– UML origins Booch Method91, 93; OMT-1 and –2 and
OOSE (use case)
– 1994, Rational Software Corporation, combining
Booch, OMT and OOSE into UML.

Concept of UML
– Core view: 4+1;
• 4: Logical + Implementation + Process + Deployment
• 1: Use Case
• Figure presents with
Fall, 2006
All Right Reserved, Zhong YAO, School of E&M, BHU
8-38
8.9. UML Technology for DSS
End Users
Programmers
Functionality
Software Management
Logical
Implementation
View
View
Analysts/Testers
Behavior
Use Case
View
Process
Deployment
View
View
System Integrators
System Engineering
Performance, Scalability,
Throughput
System Topology, Delivery
Installation, Communication
Fall, 2006
All Right Reserved, Zhong YAO, School of E&M, BHU
8-39
8.9. UML Technology for DSS

9 standard Diagrams
– Static Views
•
•
•
•
•
Use Case Diagram
Class Diagram
Object Diagram
Component Diagram
Deployment Diagram
– Dynamic Views
•
•
•
•
Fall, 2006
Sequence Diagram
Collaboration Diagram
State Chart
Activity diagram
All Right Reserved, Zhong YAO, School of E&M, BHU
8-40
8.9. UML Technology for DSS

UML module
– UML is Specifying(规范化), Visualization(可视化),
Documentation(文档化), and Constructing(结构化)
software module language, its focus is on becoming a
standard modeling language not on standard program
language.
– Common metamodel(元模型) and Notation
– UML is best for:
•
•
•
•
•
•
Fall, 2006
OOT development
Component-based development
Higher Visualization Requirement
Components Reuse
Assisting Evaluation, statistical operating flows
Easy implementation
All Right Reserved, Zhong YAO, School of E&M, BHU
8-41
8.10. Developing DSS: Putting
the System Together


Development tools and generators
– Use of highly automated tools
– Use of prefabricated pieces
– Both increase the developer’s productivity
DSS Development System Includes
– Request (query) handler
– System analysis and design facility
– Dialog management system
– Report generator
– Graphics generator
Fall, 2006
All Right Reserved, Zhong YAO, School of E&M, BHU
8-42
8.10. Developing DSS: Putting the
System Together

Fall, 2006
– Source code manager
– Model base management system
– Knowledge-base (management) system
– Object-oriented tools
– Standard statistical and management science tools
– Special modeling tools
– Programming languages
– Document imaging tools
DSS Development System Components
– Some may be integrated into a DSS generator
– Others may be added as needed
All Right Reserved, Zhong YAO, School of E&M, BHU
8-43
8.10. Developing DSS: Putting the
System Together
– Components used to build a new DSS
– Core of system includes development language or
DSS generator
– Construction by combining programming modules
– Windows environment handles the interface
Fall, 2006
All Right Reserved, Zhong YAO, School of E&M, BHU
8-44
8.11 Hardware & Software
Selection


Hardware Selection
– PCs
– Unix workstations
– Network of Unix workstations
– Web servers
– Mainframes
– Typically use existing hardware
Software Selection (Complex because)
– At start, information requirements, etc. are unknown
– Hundreds of packages
– Software updated rapidly
– Price changes
Fall, 2006
All Right Reserved, Zhong YAO, School of E&M, BHU
8-45
8.11 Hardware & Software
Selection
–
–
–
–
–
–
–

Many people involved in decision
Language capability problems
Different tools might be needed
Many criteria
Technical, functional, end-user, and managerial issues
Inaccurate published software reviews
Might prefer a single vendor
Maybe use the AHP!!!
Fall, 2006
All Right Reserved, Zhong YAO, School of E&M, BHU
8-46
8.12 DSS Technology Levels and Tools




Three Levels of DSS Technology
– Specific DSS [the application]
– DSS integrated tools (generators) [Excel]
– DSS primary tools [programming languages]
Plus
– DSS integrated tools
Now all with Web hooks and easy GUI interfaces
Relationships among the three levels (Figure 6.5)
Fall, 2006
All Right Reserved, Zhong YAO, School of E&M, BHU
8-47
8.12 DSS Technology Levels and Tools
Specific DSS
DSS Generators
(Spreadsheets, …)
DSS Tools (Languages, …)
Fall, 2006
All Right Reserved, Zhong YAO, School of E&M, BHU
8-48
8.13 DSS Research Directions and
The DSS of the Future










Fall, 2006
More AI
Faster, more powerful computers
The Web - interfaces and DB and model access
More and better GSS
ERM/ERP
Knowledge management
Better GUI
Better telecommunications
More research on theories
More research on methods
All Right Reserved, Zhong YAO, School of E&M, BHU
8-49
8.14 Supplement of Development
Method

ROMC system analysis method: Aims Mainly:
– The decision maker requirements and
– Capabilities of the DSS.
 Concept: Representation, Operations, Memory Aids, and
Control Mechanisms
 Representation: the user interface for the users of the DSS,
(the displaying forms of the information required for the
decision making, because any decision making needs lots of
the information (such as charts, images, figures and
equations))
 Operations: manipulating the various information,
including database retrieve, data sorting and calculating,
plotting;
Fall, 2006
All Right Reserved, Zhong YAO, School of E&M, BHU
8-50
8.14 Supplement of Development
Method

of course, various of modeling calculations are also included.
Memory Aids: mediate results need to store so that later
models or calculation to reuse. Therefore, Memory Aids
design is the data structure and database design. It stores the
variety of the valuable information and mediate results so that
sequential operations to use. For example:
– Recording internal and external data
– Storing temporal results produced by an ad hoc analysis
– Reminding the decision makers for some operations
backup design
– Directory of the status of information and the set values
internally
Fall, 2006
All Right Reserved, Zhong YAO, School of E&M, BHU
8-51
8.14 Supplement of Development
Method


Control Mechanisms: All of above three aspects can support
various decision process and status. Control Mechanism as a
decision procedure combined uses the representation,
operations and memory aids in which it depends on the styles
, skills, and knowledge of decision maker.
There are two kinds of functions:
– Building specification of using procedures and skills, e.g.,
using menu to select a manipulation, editing etc.
– Online Help
ROMC analyzing procedure
– Identifying various activities in each decision stage
For example, intelligence, design, choice, and implement.
Fall, 2006
All Right Reserved, Zhong YAO, School of E&M, BHU
8-52
8.14 Supplement of Development
Method
– Identifying the supporting functions for each activity.
– Building the key components of ROMC (identify the
boundaries of the ROMC)


System design
Application Case (Liang, textbook- a case of buying
a auto for personal)
Fall, 2006
All Right Reserved, Zhong YAO, School of E&M, BHU
8-53
Assignments (individual)

Interpreting following concepts:
– SDLC( or Describe the major phases of the
traditional SDLC. )
– Prototyping
– DSS teams or End-user Development
– OODM
– UML technology
– ROMC
– List eight ways in which companies speed up
application development.
– Describe the features of Extreme Programming
Fall, 2006
All Right Reserved, Zhong YAO, School of E&M, BHU
8-54