The Components of An Architecture for DSS

1
The Components of An Architecture for DSS
2
(D,D,M) Paradigm

• Dialog (D)
• The interface between users and the system.
• Data (D)
• The data base and database management system that
  support the required data and information.
• Models (M)
• The model base and model base management system
  that provide the analysis capacities.

3
The Components of a DSS
4
The Dialog Component

• Knowledge Base
• What knowledge the user must bring to the system
  in order to interact with it in dealing with the
  problem area or making the necessary decisions.
• What the user knows about the decision and about
  how to use the DSS.
• Action Language
• The option for directing the systems actions.
• Question-answer, menu-oriented, command language
  approaches, visual oriented interfaces, voice
  input (speech recognition), and so on.
• Presentation Language
• The alternative presentations of the systems
  responses.
• Text/Numbers, graphics, animation, voice output,
  and so on.

5
The Data Component

• Internal information
• Entities employees, customers, parts, machines,
  and so on.
• Concepts ideas, thoughts, and opinions.
• External Information
• Government data, public database, and so on.
• DBMS

6
The Model Component

• Types of Models
• Optimization Model ltgt Descriptive Model
• Probabilistic Model ltgt Deterministic Model
• Customer-built Model ltgt Ready-built Model
• Model Base
• Strategic Models the policies that govern the
  acquisition, use, and disposition of resources.
• Tactical Models financial planning, worker
  requirements planning, and so on.
• Operational Models production scheduling,
  inventory control, and so on.
• Model-building Blocks and Subroutines

7
Systems

• - A collection of objects such as people,
  resources, concepts, and procedures intended to
  perform an identifiable function or to serve a
  goal.
• The Structure of a System
• Closed Vs Open Systems
• Black Box
• A special type of closed system, in which inputs
  and outputs are well defined but the process
  itself is not specified.
• Effectiveness doing the right thing
• The degree to which goals are achieved.
• Efficiency doing the thing right
• A measure of the use of inputs (or resources) to
  achieve results.

8
(No Transcript)
9
Models

• - Simplified representation or abstraction of
  reality.
• Simplified
• Representative
• Keep relevant to the specific problem
• Types of Models
• Iconic (Scale) Models ex. physical replica
• Analog Models ex. graphics, simulation/animation
• Mathematical (Quantitative) Models

10
(No Transcript)
11
Modeling and Model Management

• Types of Models
• Statistical/Regression Analysis
• Financial
• Optimization
• Some Aspects
• Identification of the Problem and Environmental
  Analysis
• Identification of the Variables
• Forecasting
• Models
• Which to include, too many, or not enough?
• Model Management

12
(No Transcript)
13
Optimization

• Mathematical Programming
• allocate scarce resources among various
  activities to optimize a measurable goal
• Linear Programming
• Decision Variables
• Objective Function
• Optimization
• Coefficients of the Objective Function
• Constraints
• Input-Output Coefficients
• Capacities

14
(No Transcript)
15
Simulation

• Advantages and Disadvantages
• The Process of Simulation
• Types of Simulation
• Probabilistic Simulation
• Discrete
• Continuous
• Time Dependent Vs. Time Independent
• Visual Simulation
• Simulation Experimentation

16
Heuristic Programming

• The approach of employing heuristics to arrive at
  feasible and good enough solutions to some
  complex problems.
• Good Enough the range of 90-99 of the true
  optimal solution
• Methodology
• When to use heuristics
• Advantages and Disadvantages

17
(No Transcript)
18
Decision Making

• - A process of choosing among alternative courses
  of action for the purpose of attaining a goal or
  goals.
• - Synonymous with the whole process of management
  (involves a series of decisions, i.e., What,
  When, How, Where, By Whom, ...)
• The Phases of the Decision Process
• Intelligence
• Design
• Choice
• Implementation
• Decision Making Vs Problem Solving

19
(No Transcript)
20
(No Transcript)
21
Decision Analysis of Few Alternatives

• Decision Tables
• Maximax
• Maximin
• Equal Likelihood
• Hurwicz
• Minimax Regret
• EOL/EVPI
• Decision Tree

22
The Payoff Table

• A payoff table is a means of organizing a
  decision situation, including the payoffs from
  different decisions given the various states of
  nature.
• A state of nature is an actual event that may
  occur in the future.

23
Example - the Real Estate Investments

• Dominant Decision a decision has better payoff
  than another in all possible states of nature.

24
Maximax Criterion

• The maximum of the maximum payoffs

25
Maximin Criterion

• The maximum of the minimum payoffs

26
The Regret Table

• The difference between the payoff from the best
  decision and all other decision payoffs.

27
Minimax Regret Criterion

• The minimum of the maximum regret.

28
Equal Likelihood Criterion

• The maximum of the expected payoffs based on the
  equal probability of the states of nature.
• Ex. 50,000(0.5)30,000(0.5) 40,000

29
Hurwicz Criterion

• The coefficient of optimism, a, is a measure of
  the decision makers optimism.
• Multiplies the best payoff by a and the worst
  payoff by 1-a for each decision, and the best
  result is selected.

30
Expected Value Criterion

• The maximum of the expected payoffs based on the
  given probability for the states of nature.
• EV(office) 100,000(0.6)-40,000(0.4) 44,000

31
The Regret Table with Probabilities

• Calculate the opportunity loss and choose the
  minimum.
• EOL(office) 0(0.6)70,000(0.4) 28,000

32
EVPI - expected value of the perfect information

• Given perfect information, the expected payoffs
  would be 100,000(0.6)30,000(0.4) 72,000.
• Without perfect information, EV(office)
  100,000(0.6)-40,000(0.4) 44,000
• EVPI(office) 72,000 - 44,000 28,000, the
  amount of money one would pay for the perfect
  information.
• EVPI(office) EOL(office)

33
Decision Tree for the Example

• A decision tree is a diagram consisting of square
  decision nodes, circle probability nodes, and
  branches representing decision alternatives.

34
Decision Tree for the Example with Expected Values
35
Sequential Decision Tree
2,000,000
3,000,000
700,000
2,300,000
1,000,000
36
Sequential Decision Tree with Nodal Expected
Values
2,000,000
3,000,000
700,000
2,300,000
1,000,000
37
Analytic Hierarchy Process

• A multiobjective multicriteria decision-making
  approach which employs a pairwise comparison
  procedure to arrive at a scale of preferences
  among sets of alternatives.

38
Why AHP ?

• Deriving Weights (Priorities) for a set of
  activities according to importance.
• Multiple Criteria
• Multiple Objectives
• A single overall priority for all activities
• Tradeoffs, Fuzziness, and no unified scale of
  measurement

39
Assumption of AHP

• The methods we use to pursue knowledge, to
  predict, and to control our world are relative,
  and that the goal that we seek, i.e., knowledge,
  is itself relative.
• It admits inconsistency (including lack of
  transitivity) and measures the effect of
  different levels of consistency on the results we
  seek.
• Perceived constraints must be examined and not
  taken for granted - the only hope we have to plan
  our way out of difficult problems

40
AHP

• Causal Processes
• an action is described as an event with
  particular outcomes.
• Cause -gt Event (Outcome)
• Purposive Action Processes
• Action -gt Event (outcome) -gt Consequences
• The actor makes his choice of actions through his
  perception of the consequences that the outcomes
  will have for him.
• The AHP synthesizes these two approaches by
  identifying the outcomes that are more beneficial
  to the actors, and at the same time provides a
  way of accessing the factors (causes) which may
  have more to do with certain types of outcomes.

41
Applications of the AHP (1/2)

• Setting Priorities
• Generating a Set of Alternatives
• Choosing a Best Policy Alternative
• Determining Requirements
• Making Decisions Using Benefits and Costs
• Allocating Resources
• Predicting Outcomes (Time Dependence) - Risk
  Assessment

42
Applications of the AHP (2/2)

• Measuring Performance
• Designing a System
• Ensuring System Stability
• Optimizing
• Planning
• Conflict Resolution

43
Procedure for AHP Analysis

• Determining the requirements of the system
• What do we need to do?
• Generating alternatives to satisfy those
  requirements
• What are the possible ways of action?
• Setting priorities according to the importance of
  the requirements in order to implement the
  alternatives to attain some higher objective
• Choosing the best policy alternative, or a mix of
  the best policy alternatives
• Using forward and backward projections to obtain
  a stable outcome.

44
The Weighting Matrix

• Eigenvalue and Eigenvector problem
• AX lX, where l n.
• ?Wi 1

45
Rating Scale for Comparison
46
Example - Buying a Car
47
Pair Comparison (1/2)
48
Pair Comparison (2/2)
49
Composite Priority
50
Criteria User Interface