UNIT-II

DSS Architecture

Information system architecture

The architecture of an information system refers to the way its pieces are laid out, what
types of tasks are allocated to each piece, how the pieces interact with each other, and how
they interact with the outside world.

DSS or IS Architecture has to achieve Properties

 Interoperability of systems, so that information can be brought to the point of use

quickly and easily.
 Compatibility of systems, so that resources can be shared easily and leveraged across
the organization.
 Expandability of systems, so that limited single – function components do not create
bottlenecks that obstruct the growth of the organization.
Significant Benefits of DSS Architecture

Non Technical benefits:

 the ability to create a common vision that keeps all project participants working in
tandem,
 the ability to communicate system concepts to management,
 The ability to communicate needs to potential vendors.
 The ability of other groups to implement systems that must work with the DSS.

Technical benefits:

Technical benefits of a DSS architecture include the ability to plan systems in an effective
and coordinated fashion and to evaluate technology options within a context of how they will
work rather than abstractly.

The 8 Essential Elements of DSS Architecture

 Its database or databases, including any existing data bases, internal or external to the
organization and any databases that are created specifically for DSS use.

1
 Its model or models, including information about their sources of data, the
organizational responsibility for maintaining them and limits on access on them.
Models may not be practical but may be define major categories
 Its users including any assumptions about their locations, jobs, levels of education,
and any other factor that may affect their use of the DSS.
 Software tools through which the users access the database and the models.
 Software tools through which system administrators manage the database and the
models.
 Hardware and operating system platforms, at a generic level, on which the data base
and models reside, on which the programs run, and through which users access the
DSS.
 Networking and communication capabilities through which these platforms are
interconnected. These must reflect individual needs to connect to one or more servers
and databases.
 The culture of an organization that will use the DSS. If its culture is centralized, it
may be acceptable to have a central database.

2
The major DSS platforms/ hardware environments are:

The central corporate system

The central system linked to other computers on users desktops via a network.
A separate system that obtains data from the central system and provides it to users.
A free-standing system at the users desk.
A combination of the above.

Hardware Tools

DSS on the Central Corporate System

• Corporate applications are generally put on the central system.

• These systems run operational transactional based applications such as

accounting, order entry, and time billing.

• And also often run strategic applications such as EDI with customers
or facilities for rapid inquiry into the status of critical orders.

Such central, multi-user systems usually use an integrated DBMS for the organization's
important information.

Given the central shared system already exists; it is tempting to put new applications on it. –
Which is monitored by the Capacity Planners.

3
When corporate applications are generally put on the central system, and if this mode of
regular load increases a new DSS can be put up.

Doing so has both advantages and disadvantages.

Advantages:

– MIS staff members are already familiar with the system.

– Many users already know how the system works.

– The necessary hardware to use the system is already in place in the

organization.

– DSS applications can access the central database directly for up to date
information.

– The processing power of a large system can run complex DSS, such as large
simulation models.

Most large systems have a variety of available DSS tools.

Disadvantages:

– The central system was probably not originally selected for DSS applications
and may not be well suited to them.

– The cost of additional resources to run a DSS may exceed the cost of smaller
computers to handle the same task.

– There may be a tendency to force unsuitable software tools into DSS use,
rather than getting suitable ones, because they are already there.

– DSS responsiveness may be impacted by other loads on the systems.

– DSS users may have to wait in line behind other application development
efforts.

– The user interface of most large systems is not as easy for non technical
people to use as that of most smaller systems.

– Support for graphics may be poor, even if an application can handle graphics.

4
5
DSS and Client/Server Computing

 The approach of client/ server computing allows a shared computer to provide

decision support data and handle other aspects of the tasks to which it is well
suited, while allowing more economical desk top computers to provide the
user interface, while the server does the rest of the job.

 In other words, many organizations deal with the aforementioned

disadvantages by providing individual users, via their desktop computers, with
access to data on the central computer.

 The desktop computers then handle the computations and other processes of
DSS.

 This approach is called as a Client/Server computing.

 The system storing the database is called Server.

Fat client:

 Client /Server approaches where the client does most of the work,
relying on the server only to supply it with raw data, are often
referred to as fat client approaches.
Thin client:

6
  A client system that does little more than provide a user interface to
results provided by the server is often called thin client.
Network computers:

 Is a desk top system that can access the web but has no stand alone computing
capability.
Intranet:

o An intranet is an internal network that uses web protocols to control data

transfer.
Major factors of C/ S Computing

 The Computing Power Relationship between two Systems

 The load Pattern

 Minimizing Communication Needs

Advantages & Disadvantages using C/S Model

Adv

 Cost Saving

 H/W & S/W Optimization

 Modular Nature of Expunction

Dis Adv

 Security Problems

 Complexity of Application

 Addition people & H/W

The Internet and Client/ Server Computing in DSS

 The world wide web can also be used to supply decision support data to users

 The web architecture,

7
  The HTML – for developing web pages

 The Java language – that most browsers can interpret,

 The Java Script language – for commands added to HTML, and

 The standards that web browsers must follow define, in effect, a

platform for applications.

 Here we can talk about – Intranet and Network computer (NC)

 It has its disadvantages

 Web access can be slow especially over standard modems and telephone lines

 Web site designs and DSS designs are separate skills. Both are required for an
effective web based DSS. Serving a given number of users with a web based
or web like application requires a more powerful server.

DSS Using Shared Data on a Separate System

• It is not necessary for the server in a client/server system to be the central corporate
computer that stores the live, operational database.

• It is often a good idea to extract meaningful decision support data from the
operational database and load it into a different computer.

• That 2nd computer can then be accessed directly via terminals, or can act as a server in
the client/server computing model.

– The new system can range from microcomputer to a “supermini”.

– The choice depends up the required capacity, the required software tools, and
the number of concurrent users.

– The linked system approach is effective when the application allows for a DS
database that is separate from the firms TPS.

– As we know, such a database is often called a DATA

WAREHOUSE

Define a data warehouse

8
 – A data warehouse is a read only database.

– Its users cannot update the “live” or organizational information.

– That limits the use of data warehouse to situations where decision makers
access corporate information but do not change it.

– If an application running in this environment must update the main database, it

must be done in some other way.

– One approach is to have DSS computer mimic a transaction processing user

and submit transactions.

• Advantages:

– The DSS hardware can be optimized for the specific purpose.

– DSS hardware need not be shared with other applications.

– The cost of upgrading the central corporate system to run DSS applications is
avoided.

– Compared to full blown client/server computing, in which the applications

can update the server database, the unidirectional data warehouse environment
is simpler and less prone to errors.

– Most servers and micro computers are user friendly or can be made so by an
available graphical user interface package.

• Disadvantages:

– The need to transfer data between two systems, to create the data warehouse
and keep it up to date.

– The MIS staff must be familiar with two systems, users who access both
systems may need two types of terminals.

– Decision support data are only as current as the most recent down load to data
warehouse.

9
 – Re-creating the data warehouse may take several hours, and the DSS system
may not be available for use during this time.

DSS on a Stand - Alone System

 Many DSS that do not access a large central database are candidates for stand alone
systems.

 Some run models whose input, consisting of only a few numbers, comes
entirely from its users knowledge or from external sources and can be keyed in
easily for each run.

 DSS that do not access a large central database are typically model-oriented or
process-oriented.

 A stand-alone DSS can be run on a computer dedicated to the DSS task or on a

multiple-user computer used in time-sharing mode.

Advantages:

 The system can be totally optimized for DSS.

 The complexity of sharing a resource with other users is avoided.

 The complexity of integrating a two system application is avoided.

 There is no overhead in communicating with other systems.

 The person or department owning the system has complete control over it.

Disadvantages:

 Any data the system requires must be provided by its user.

 Sharing the results with other users, or using a shared database, may be more
difficult that it would be if the system were designed to operate in a shared –
data environment.

 The individual user or department responsible for the stand – alone system
must usually deal with its own system administration needs.

10
  It may be difficult to integrate a stand – alone system with corporate
applications at a later date if no thought was given to compatibility in the
planning stages.

Open Systems and DSS

• What is an Open system?

– Open systems are systems whose interfaces are not under the control of any
single hardware or software supplier.

– Rather they are defined by neutral bodies through an open process in which all
parties can, if they choose, participate.

– Once they are agreed upon, both the interface specifications and the right to
produce systems that conform to those specifications are available on equal,
and equitable terms to all firms.

Aspects of openness

– System openness encompasses multiple aspects.

• The first aspect is system interconnection, or networking.

• The second aspect of openness is the way individual computers

operate.

• The multi user open operating systems are UNIX, POSIX, and
at the desk top level, Microsoft windows.

• The third aspect of openness is hardware interface openness. Hardware

interfaces to many microcomputers and minicomputers are in the
public domain or can be licensed.

• DSS has accelerated the growth of open systems. The reason: the arrival of DSS
injected a second component into what had been, historically, single purpose
transaction processing systems.

Choosing a DSS Hardware Environment

• Questions to be asked before choosing a right DSS hardware approach.

11
 • Are there any corporate policies that you must follow?

• How large and widespread will the DSS user community be?

• Are most of the prospective users already using a particular system?

• Is there any corporate mainframe with sufficient capacity? & Is another

powerful server system available?

• Do prospective users already have microcomputers or work stations that can

handle the application?

• If new system is required, will the existing central system be able to share
data with them?

• Do the necessary development tools, if any, are the prospective DSS

developers already familiar? & does the application require access to a
database?

• Does the application’s use off the corporate database require only the ability to
read the data, or must it also be able update the data?

• How much processing power does the application require?

• Are prospective users capable of performing basic system administration

tasks, such as installing software and backing up data files?

Software Tools

DSS Software Categories

There are four fundamental ways to obtain any software capability

 To purchase a turnkey package,

 To customize a package,

 To use specialized tools or “generators” designed for the task at hand,

 to write the necessary programs from scratch.

12
  Package software is couple of programs combined together which is recommended by
the company

 Custom software is where you can select the kind of programs and the execution of
the software according to your need

 The choice between packages (customized or not) and custom software usually
depends on two factors.

 The degree to which your needs resemble those of many other organizations.

 The financial impact of the application, which determines the value of getting exactly
the capability you want versus what you can get in a standard package.

Standard Packages

Standard packages to help make specific decisions have been developed for a few common
decisions.

 The characteristics of this decision that make it attractive to package suppliers are:

 Many people make these decisions, so the potential market is large.

13
  The decision has financial importance to many of these potential users.

 The underlying factors are the same for all of them.(individual investors vary
in the importance they attach to different performance measures and in the set
of stocks of potential interest, but a price-volume chart is a price-volume chart
to everyone.)

 The support infrastructure – public databases that can provide the necessary
data to anyone, free or for a small fee – already exists.

 Decisions having these characteristics are, unfortunately, rare and especially so in the
business world.

 Where they exists, and where companies have developed DSS to help with
them, these companies often do not want to give away their secrets to their
competitors.

Specialized tools and Generators

14
  DSS tools & DSS generators allow DSS developers to utilize standardized “building
blocks”, which support these common features, to develop their own custom
applications.

 The terms DSS tools and DSS generators both quite common in the DSS literature,
create a great deal of confusion.

 Sprague was 1st to use these terms.

 DSS tools would generally be used to create DSS generators, which in turn
would generate the specific DSS used by decision makers.

 The use of DSS generators is associated with the higher level, more business-
oriented, roles.

 The major categories of specialized software used to assist DSS development are

1) Database management packages

2) Information retrieval (query and reporting) packages.

3) Specialized modeling packages (including spreadsheets) and

languages.

4) Statistical data analysis packages

5) Forecasting packages

6) Graphing packages.

Database Management Systems

15
  A data base management system allows you to store data in an organized form and
retrieve it in on the basis of specified selection criteria.

 The four basic database structures:

A hierarchical database

 Manager links records of different types in a strict hierarchy from top

to bottom. Each record , except those at the top level, is associated with
a specific parent record. It is weaker in handling more complex
relationships, such as many to many relationships.

A network database structure

 More flexibility in the way different files are linked. The use of the
term network here is to describe a database structure. Compare to
hierarchical DBMS, network DBMS software is more complicated
because it must be able to represent more complex data structures.
Many network DBMS follow the CODASYL standard.

A relational DBMS

 Stores its data in the form of two dimensional tables. The database
management software links record from several tables by matching
corresponding fields. Relational databases provide a good foundation
for a data ware house designed to respond to unplanned queries. The
advantage of relational DBMS is interface standardization.

A multi dimensional DBMS

 Is particularly useful in data warehouses.

16
Major objects in Database

 Data Managers or File Managers

 Entity

 Attribute

 E-R Diagrams

 Conceptual Data models

 Keys and Key attributes

Information Retrieval Packages

 Most informational retrieval packages are designed to pull user specific data out of a
file or a database.

 You might type the computer equivalent of “get me our sales history for
Acme. Inc”.

 The database will respond with the requested data – or with “I don’t
understand what you what” or “I don’t have that data”.

 SQL & other

 A different type of package finds information by searching for words

in a text database. Ex: library database / world wide web.

 Systems that retrieve few irrelevant items are said to have high precision, and systems
that miss few relevant items are said to have high recall.

17
Specialized Modeling Languages

 What are the advantages and disadvantages of using spreadsheets for business
planning?

 Many DSS incorporate models of various types. The packages do not

incorporate any models themselves. They simply make it easier for the user to
define the characteristics of his or her model to the computer.

 Taking the example of a spreadsheet which is a standard tool, which most

knowledge workers and managers manipulate accounting models.

 A spreadsheet, in its basic form, looks like a two dimensional grid of cells.
They can create several kinds of graphs automatically from the data they
calculate. They may extend two dimensional grid to three or more dimensions.

Statistical Data Analysis Packages

 What do statistic packages do ?

18
  By determining how a system has reacted to certain factors in the past, we
hope to estimate how it will react in the future and thus gauge the likely
outcome of our decisions.

 These decisions are often statistical.

 Most statistical packages allow a decision maker to deal with more than one
dependant variable at a time. Statistics packages can create cross tabulations
to analyze data. It allows decision makers to identify clusters in data.

 A common management application of statistical data analysis is in statistical

quality control.

Forecasting Packages

 Forecasting can be defined as predicting the future based on facts known at the
present.

 One type of forecast is a pure prediction of a phenomenon that will take place
in the future.

 Another type of forecast is derived from a model of how known factors

influence other factors that cannot be predicted as easily.

 The basic steps of time series forecasting/analysis.

 Time series analysis extrapolates a trend solely on its historical changes with
no reference to any underlying mechanism that may be responsible for that
trend or to a possible relationship between the forecasted item and some other
variable.

 Extrapolation takes the available data, fits a curve to them and extends the
curve into the future.

 All forecast can be dangerous to trust the output of a forecasting package too much sts
assume that nothing of importance is going to change from the base period to the
period being forecasted.

 Developing a time series forecast involves three steps:

19
  Determine the periodic cycles of the phenomenon.

 Some phenomena react to more than one cycle: perhaps a weekly

cycle, a seasonal cycle, plus some sensitivity to overall business
cycles. In this case, it is necessary to identify all the major cycles to
which the phenomenon is sensitive.

 Determine the secular trend of the phenomenon.

 The secular trend can often be found, or at least visualized, by

averaging several periodical cycles of a phenomenon.

Graphing Packages

 Most people assimilate data most readily in the form of picture.

 Most decisions are based on trends or differences that are clearly apparent on a graph.

 Most spreadsheet, statistical and forecasting packages can produce graphic

output.

 Yet their basic purpose is not graphics, so their capabilities are often limited in
this regard.

 Specialized graphing packages can take their output, in number of

standardized interchange formats, and create a variety of charts and graphs
with it than the original package could.

 The need for graphing is dropping as the graphing capabilities of other types
of software improve.

Programming Languages

 Third generation programming languages.

 Third generation languages have the advantage of standardization.

 Such as Pascal and C and even the older COBAL, fall under this category.

 3GLs have refined codes that utilizes hardware efficiently.

 High volume transaction processing programs fall under this category.

20
  3GL programming is labor intensive, time consuming and error prone.

 Fourth generation programming languages.

 The fourth generation languages specifies what the computer should do and
not how the computer is to do it.

 A 4GL like FOCUS uses flexibility and run time efficiency for speed of
development.

 The need to interface to one or more existing databases and to modules written
in a 3GL are often important factors in 4GL choice.

 4GLs are not widely accepted because:

 3GLs are designed primarily for applications such as summary reports not
complex data manipulation.
 3GLs are not standardized.
 Most existing mainframe based transactions are written in 3GL.
 The cost of the 4GL and its supporting software is to high.
DSS User Interfaces

 Planning the user interface is an important part of planning any DSS, perhaps more so
than with most other types of information systems.

 DSS users have more sporadic and more varied needs than do most other IS users, and
their activities tend to have higher organizational impact.

 Factors to be considered in planning the DSS user interface include,

 minimizing task performance time,

 minimizing learning and recall time,

 maximizing system versatility,

 minimizing errors and

 Adapting to different user needs, including the availability of help for those
who need it.

 Additional factors to take into account are

21
  the users ability to keep several thing in mind, i.e., adaptability,

 Concentration,

 user fatigue,

 Uniformity, and fun, and

 command consistency with other systems with which the users are familiar.

 Why you should not rely exclusively on color to convey information to DSS user?

 Color can often enhance a user interface but must used with due consideration
for those users who cannot see a difference among the developers color
choices.

User Interface Styles

 There are four general ways to control computers are Command line interfaces,
Graphical interfaces, Menus, and Question and Answer dialogues.

 Animation is an increasingly popular part of DSS interfaces.

 Each of these applies both to the DSS itself and to the operating system under
which the DSS runs.

 Developing an application that conforms fully to the standards of a GUI is,

despite the availability of operating system facilities, more difficult than
developing a comparable application that operates via menus and typed in
commands.

 Command-line interfaces:

 Are the oldest form of computer control, going back to the days in which each
command was entered on a punched card.

 They still underlie several operating systems, such as the complex job control
languages of mainframe systems.

 It is the users responsibility to know what commands are available and how to
enter them with their parameters.

22
  Graphical user interfaces:

 Dominate at the desktop level and are increasingly popular on larger systems.

 Each small picture, called an icon, represents a program or a data file.

 Mouse

 Menu bar or Action bar

 Key board short cuts

 Commands can be entered into a DSS from Menus, which the application displays on
the screen.

 These are different from GUI menus.

 Question-and-answer dialogues:

 Are appropriate when the user must specify data values or other parameters
before the system can carry out a desired operation.

 In the GUI environment, question-and-answer dialogues are usually presented

via dialogue boxes such as the one in the figure, was created by the MS
Excel….

Hypertext/Hypermedia

 What is hypertext/hypermedia?

23
  Hypertext refers to electronic document whose parts are electronically linked
to avoid limitations of traditional linear media such as bound books.

 Hypermedia extends to hypertext by adding text , graphics, animation, and

sound to electronic document.

 Together, hypertext and hypermedia allow users to access information as

required by their needs rather than by the structure of the database or the
limitation of a computer program.

 What is virtual reality?

 Virtual reality uses hardware and software to give an impression that a

computer created environment exists.

 VR is a step beyond animation.

 Immerse VR attempts to give the user a sense of inhabiting a computer created

space with the ability to manipulate objects in that space.

DSS Development Process

Development Process

Three approaches to information system development

The Traditional system development life cycle(SDLC) approach.

 The prototyping approach.

 End User development approach.

24
25
26
Alternative Development Methodologies

 Parallel development

 Rapid application development (RAD) methodologies

 Phased development

 Prototyping

 Throwaway prototyping

Parallel Development

 Multiple copies of design and implementation phases

 To develop separate subsystems

 All come together in a single implementation phase

27
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

28
Prototyping

 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

29
Throwaway Prototyping

 Like prototyping and SDLC

 Analysis phase is thorough

 Design prototypes assist in understanding the system

30
Prototyping for DSS Development

 Problems are semi structured or unstructured

 Managers and developers may not completely understand problem

 Use prototyping

Prototyping Terms

 Iterative design

 Evolutionary development

 Middle-out process

 Adaptive design

 Incremental design

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

Advantages of Prototyping

 Short development time

 Short user reaction time

 Improved user understanding

 Low cost

Disadvantages of Prototyping

Gains may be lost in

31
  Thorough understanding IS’s benefits and costs

 Detailed description of information needs

 Easy to maintain IS design

 Well-tested IS

 Well-prepared users

End-User-Developed Systems

 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

User-Developed DSS Advantages

1. Short delivery time

2. Eliminate extensive and formal user requirements specifications

3. Reduce some DSS implementation problems

4. Low cost

User-Developed DSS Risks

1.Poor Quality

32
2. Quality Risks

 Substandard or inappropriate tools and facilities

 Development process risks

 Data management risks

3. Increased Security Risks

4. Problems from Lack of Documentation and Maintenance Procedures

DSS Development Project Participants

 Users

Non-technical people they will use DSS for solving the problems in functional areas
of business

 Intermediary

The people who helps to Users and provide training to them mainly how to use DSS

 DSS builder or facilitator

The people who will take decision regarding H/W, S/W, database incorporated with
DSS

 Technical Support Persons

This is the programmers who integrates existing packages into one overall system and
carries out custom programming that contributes directly DSS functionality

 Tool smith

This person focuses on the tools that will be used in constructing the DSS and the
packages that will be combined into it she/he is an expert on those tools.

The Implementation Stage

 Obtaining and installing the DSS hardware

33
  Installing the DSS and making it run on its intended hardware

 Providing user access to the system

 Creating and updating the database

 Training the users on the new system

 Documenting the system for its users and for those who will be responsible for
maintaining it in the future

 Making arrangements to support the users as the system is used

 Transferring ongoing responsibility for the system from its developers to the
operations or maintenance part of the MIS group

 Switching over to the new system from previous methods of making decisions or
obtaining information to make them

 Evaluating the operation and use of the system

System Conversion

 In direct conversion or direct cutover

 Parallel conversion

 Pilot conversion

 Phased conversion

Overcoming Resistance to Change

 Change is inherit in the implementation of any new system

 The Perspective Implementer must aware of personal and Organizational impact of

change

 Managing change is not a technical issue.

 Organization culture is a key consideration in planning any change

 Change can be thought of as taking place in Three Stage process

34
Unfreezing:

Creates the conditions and attitudes that are necessary before meaningful change can
take place.

Moving:

constitutes the change itself: putting new system into operation to replace

the old version

Refreezing

Involves making the new system as much a part of the organizational fabric as the old one

Institutional DSS:

 intended for one-time use

 It does need level of acceptance.

Adhoc DSS:

 intended for one-time use and discarded as soon as the need for it goes away.

 It does not need level of acceptance.

Unfreezing

 The Unfreezing stage it is important to create a strong motivation for change

 Present system felt un satisfactory then we will move to new one.

 Create common vision to accept it

 Clarify technical & non technical benefits

 Clarify organization & employee benefits and how it improves decision making.

Moving

 The most visible component of moving stage is the actual conversion from old
decision making methods.

35
  Pilot Conversion is to be used (training)

 Parall conversion take place (where max difference b/w old and new system)

Refreezing

 User of the new system have the necessary internal commitment to using it on going
basis.

 Allow sufficient time for each stage of change to take place

 Make sure each stage of change is successful before moving to the next.

 Also found that the ability to reverse the change.

DSS Implementation Issues

 Unfamiliarity with this type of system

 Response Time

 Reliability and availability

 Poor data quality

User related DSS Implementation Issues

 Voluntary or Mandatory use

 User and Management support

 Unstable user community

 Response Time

 Training and documentation

 An Estimate of completion time before an operation starts

 System Load is expected to increase and thus slow response times in the future

 Training

36
  Availability of Support

Ethical Issues in DSS Implementation

 Storage of Information

 Use of Information

 Shared of Information

 Human Judgment

 Combining Information

 Error Detection and Correction

Models in DSS

 Used to make prediction what would happened in the real word.

 Enable decision makers to evaluate each alternative without trying them out.

 Basic type of System Models

 Graphical Model –DFD,Map

 Narrative Model – Describe the System in Natural Language

 Physical Model – Smaller or Idealized representation of real system.

 Symbolic Model/Information Based Model – Represent Reality by Data

Types of data:

i. True/False or Yes/No (Boolean Variables)

ii. Character String

iii. Numerical Value

Graphical Model

37
Narrative model

Physical Model

38
Family Tree of Model Types

Mathematical Model

 Identify variables

 Establish equations describing their relationships

 Simplifications through assumptions

Balance model simplification and the accurate representation of reality

Systems versus process models

39
  A mathematical model is a information – based representation of an actual
system.

 system model – models the system that we wish to study

 Mathematical models are an important subset of system models

 process model (prescriptive models) – models the process that humans follow in
making a decision about a system.

 process model can be normative or descriptive.

Static versus dynamic models

 Static models show the values that system attributes take when the system is in
balance.
 The static model, by showing when a system is in balance, can tell decision makers
how the system will eventually stabilize even if it does not show them how it gets to
that point.
 And this model may also be able to provide the results more quickly than a dynamic
one, allowing decision makers to consider more options.
 For a static model to apply to a dynamic system, that system must be in equilibrium
(in balance), often referred to as being in its Steady State.
 Over a period of time, the number of cars arriving and leaving the factory.

 The Dynamic System is in which the passage of time with cause-and-effect

relationships connecting one time period to the next, is essential to system behavior.

 The Static System is in which the passage of time does not play a part.

 In a dynamic model, which can only be of a dynamic system, the flow of time is
inherent in the modeling process.

40
  Data values change over time.
 An assembly line model works this way as it tracks products through a manufacturing
process.
 When a subassembly finishes one step of process, the next step can begin.
Continuous versus discrete- event models

 Continuous-system simulation models describe physical or economic

processes in which the numbers that describe the system vary continuously.

 For example, blood pressure varies continuously over time.

 It does not jump instantaneously form one value to another.

 Rather there are processes that cause it to rise or fall at given

rates –(there are events, such as stress, that cause it to change
rapidly, but the pressure curve is still continuous.)

 Discrete – event models deal with systems in which individual events occur
at identifiable points in time and change the state of the system
instantaneously from one value to a different one.

 For example, the arrival of an order at a valve manufacturer may

change the state of the system from “2 Model 100 valves on order” to
“3 Model 100 valves on order. ”

 Socio economic planners use continuous models, because their work is not in
the mainstream of corporate DSS.

 Since discrete – event models suit most business planning needs, they are the
most common type of dynamic system model found in real DSS.

Deterministic versus stochastic models

 A model is deterministic if its outputs are fixed for a given set of inputs,
stochastic if they reflect an element uncertainty.

 A deterministic model of a 3-step assemble process might indicate that

each step takes exactly one hour, so the entire process takes 3 hours.

41
  A stochastic model of the same assembly process might have the
duration of each step vary around the one-hour average, with a normal
distribution having a standard deviation of ten minutes.

 This is more realistic, because most physical processes and

nearly all human activities have some variability from on
instance to the next.

 The stochastic models output varies randomly over a range of possible

outcomes.

 Stochastic models, by contrast, fall into what we called the representational

model category of DSS.

Discrete – Event Simulation Models

 A model is the description of the system, usually in the form of a computer program

 We call a model a simulation model when we want to make it clear that we are not
discussing some other type of model or some other use of a model.

 The Simulation is a process of using this model to study a system.

 The Simulation is a process.

 Simulation cannot exist without a suitable model, and a model meant for use in
simulation may be useless for anything else, but the concepts are different.

 The concept of discrete- event simulation:

 The model represents the state of the system by the values of data elements
(variables) in the computer.

 And the values of these variables change as events occur in the system.

 Most discrete – event simulation models are stochastic or probabilistic.

A Discrete-Event Simulation Example

 You don’t need a computer to construct a simple simulation model and run
experiments with it

42
  We want to model the operation to a Barbershop to determine how many barbers
should we hire, how many customers will arrive and what will be the service time –
based on the Die we roll number of times and then service time depending on the
same.

And we can’t appoint more than 2 Barbers.

Designing a Discrete – Event Simulation Model

 The process of designing a discrete – event simulation model. The process consists of
the following 8 steps:

1) Determine the objective of the model.

2) Define the system itself.

3) Define the state of the system in terms of a set of state variables or

uncontrollable variables.

4) Define the events that can affect the state of the system and the impact of each
event on each state variable.

5) Choose the time units which the simulation will use.

43
 6) Define, statistically, the rate at which each event occurs.

7) Determine the statistics you would like to obtain from your simulation and
what data you need in order to obtain them.

8) Define the initial (starting) state of the system.

 Having established the parameters of your model, you can now program it in a special
– purpose languages such as GPSS, SLAM or Simscript, or in a general purpose
language such as C.

 Simulation languages generally include capabilities that will allow you to develop
your model more quickly.

Complete Simulation Studies

 A full – scale simulation study runs the simulation several times for each state of
controllable variables to give us a distribution of results.

 Instead of running the model several times, we might consider one long run covering
period of time equal to the sum of the shorter individual runs.

 In general , one long run is not as good as the same total time divided into several
shorter ones. The reason is a phenomenon called autocorrelation.

 Autocorrelation means that what happens later in a run depends on what happened
previously in the same run.

 Example – supermarket.

Random Numbers, Pseudo - Random Numbers and Statistical Distributions

 The behavior of a simulation model depends on the numbers that determine when
each event occurs.

 There are many ways to generate random numbers.

 Simulation models that use random numbers are often called Monte Carlo
simulations after the casino in Monaco.

 It is possible to equip a computer with a physical random number generator.

44
  Using truly random numbers has the disadvantage that their sequence is not
repeatable.

 For the above reason, virtually all simulation models use pseudo - random
numbers.

 These are numbers generated by a repeatable formula, which behave

statistically as if they were truly random.

 All simulation packages and most programming languages have built-in

function that return uniformly distributed pseudo – random numbers over a
more useful range.

 The built-in random number generators of most systems are not

perfect.

 Computer science journals regularly critique popular ones and suggest

improvements.

 In order to obtain the value for the simulation, pseudo – random numbers are used
together with the cumulative distribution functions (CDF) of the statistical
distributions that apply to the system being modeled.

 In practice, CDF’s are usually approximated by a small number of straight – line

segments.

 Simulation packages include a set of standard CDF’s for common distributions.

 Simulation models must be exercised for long enough for fluctuations due to specific
choices of random variables to die down.

 Several runs of moderate length are generally better for this purpose than one long
run of the same aggregate length.

 When you want a non uniform distribution, you must convert the output of
the built-in function to a number from the desired distribution.

This is done via cumulative distribution function (CDF).

Static Simulation Models

45
46
 Packages are available to help decision makers use static, stochastic, simulation
models.

 The advantages are :

 It is easy to add statistically defined variability to the deterministic model.

47
  A wide variety of statistical distributions is available, many of which would be
quite difficult for a decision maker to define directly.

 The parameters of the distributions are easy to change.

 The output facilities exceed what spreadsheet packages offer for this type of
problem.

 The disadvantage are:

 There is a need to use two packages, one to define the model and to run it as a
simulation.

 More complex cases call for custom programming.

48