You are on page 1of 54

Multiple Criteria Decision Making:

from Decision Making to Decision


Support

Hsu-Shih Shih, Ph.D.

Graduate Institute of Management Sciences


Tamkang University, Tamsui
Taipei, Taiwan, ROC
Personal Background
™ Ph.D., Industrial Engineering, Kansas State
Univ.
™ Visiting Professor, Univ. of Pittsburgh
™ Major Interests
¾ Decision analysis (MODM, MADM, GDM)
¾ Decision support (DSS, GDSS, NSS)
¾ Operations research (Linear/Dynamic programming)
¾ Soft computing (Fuzzy sets, Neural networks)
™ Website
¾ http://163.13.193.161
Brief Contents
™ Introduction
™ Definitions
™ Categories of Decision Making
™ History of MCDM
™ Categories of MCDM
¾ Multiple Attribute Decision Making
¾ Multiple Objective Decision Making
¾ Group Decision Making
™ Decision Support
™ Main Concerns
™ Future Studies
Introduction
™ Why we need decisions?
¾ Decisions permeate life
™ How to make a decision?
¾ Get a big picture about decision-making
™ How to make a better decision?
¾ Obtain it through training
™ How to get a better outcome?
¾ Unable to define it through rationality
Some Basic Terms (I)
™ Decision
¾ A choice that you make about what you think
should be done or about which is the best of
various alternatives
™ Decision Making
¾ The process of or the responsibility for making
decisions
™ Decision Support
¾ An effort to aid the process of decision making
Some Basic Terms (II)
™ Decision Maker
¾ One major role of managers
™ Analyst
¾ Who helps managers describe the whole system
(or problem) not merely in terms of its parts but
most significantly by the pattern that connects
them
™ Preference
¾ The power or opportunity of choosing; the act, fact,
or principle of given advantages to some over
others
Some Basic Terms (III)
™ Decision Analysis
¾ the discipline comprising the philosophy, theory,
methodology, and professional practice necessary
to address important decisions in a formal manner
™ Multiple Criteria Decision Making
¾ the study of methods and procedures by which
concerns about multiple conflicting criteria can be
formally incorporated into the management
planning process
Categories of Decision Making
™ Rationality (optimal solution)
™ Bounded rationality (satisficing solution)
™ Behavior
¾ Politics
¾ Experience
¾ Intuition
Involved Areas (I)
™ Psychology and Behavioral Sciences
¾ Descriptive decision making
¾ Decision people make; How people decide
™ Statistics, Mathematics, and Economics
¾ Normative decision making
¾ Logically consistent decision procedures; How
people should decide

Bell et al. 1988


Involved Areas (II)
™ Operations Research and Management
Science
¾ Prescriptive decision making
¾ How to help people to make good decisions; How
to train people to make better decisions

The main concerns of the topic


Elements of MCDM Problems
™ Set of Alternatives
¾ Finite / Infinite choices
™ Set of Criteria
¾ One / Multiple criteria
™ Outcome of Each Choice
¾ Outcome can be deterministic/uncertainty
™ Preference Structures of the DM
¾ Preferences are clearly specified or not

Yu 1985
Whole Process of Management
™ Five Steps:
1) Intelligence
2) Design
3) Choice
4) Implementation
5) Monitoring

The process of problem-solving /decision


making

Simon 1977, Turban et al. 2005


Types of Problems
™ Types of Problems
¾ Structural (programmed, routine processes)
¾ Semi-structural
¾ Un-structural (non-programmed, fuzzy and
complex processes)
™ Knowledge about the Future Outcomes
¾ Certainty (deterministic)
¾ Risk (stochastic)
¾ Uncertainty (ignorant)

Turban et al. 2005


Process of Problem-solving
1. Identification of a Problem
Ø
2. Identification of Decision Criteria
Ø
3. Allocation of Weights to Criteria
Ø
4. Development of Alternatives
Ø
5. Analysis of Alternatives
Ø
6. Selection of an Alternative
Ø
7. Implementation of the Alternative
Ø
8. Evaluation of Decision’s Effectiveness
Ø
( Go To 1 )
History of MCDM (I)
™ 1951 – H.W. Kuhn, A.W. Tucker considered problems
with multiple objectives
™ 1955 – A. Charnes, W.W. Cooper, R. Ferguson
mentioned the essence of goal programming (a book
published by Charnes and Cooper in 1961)
™ 1968 – B. Contini, S. Zionts developed a multiple-
criteria negotiating model
™ 1973 – S. Zionts, J. Wallenius proposed the Zionts-
Wallenius interactive method for MODM
™ 1970s – S. Zionts, J. Wallenius, P. Korhonen worked
on methods and decision support systems for MODM

http://www.mcdmsociety.org/facts.html#History
History of MCDM (II)
™ 1959 – R. Howard, G.E. Kimball worked on
sequential decision processes for decision analysis
™ 1968 – H. Raiffa published a book on decision
analysis
™ 1976 – R. Keeney, H. Raiffa established the theory of
multiattribute value theory (MAVT)
™ 1973 – J.P. Evans, R.E. Steuer developed a multiple
criteria simplex method (M. Zeleny independently
carried out the similar work in 1972)
™ 1979 – A. Tversky, D. Kahneman made contributions
in behavioral decision theory (prospect theory)
History of MCDM (III)
™ 1960+ – B. Roy and his colleagues developed
ELECTRE, a family of multi-criteria decision analysis
methods, for outranking decision.
™ 1970+ – T.L. Saaty introduced the analytic hierarchy
process and generalize it as the analytic network
process in 1996
™ 1977 – W. Candler, R. Norton defined multi-level
programming for conflicting objectives of economic
problems
Awards from International Society on MCDM
Categories of MCDM

™ Multiple Attribute Decision Making


™ Multiple Objective Decision Making
™ Group Decision Making
Multiple Attribute Decision Making

Multiple Criteria Decision Making


Characteristics of MADM
™ Multiple Attributes/Objectives
™ Conflict among Criteria
™ Incommensurable Units
™ Distinctive Alternatives

Hwang and Yoon 1981


Methods for MADM (type of information
from DM)

™ No Preference Information Given


¾ Dominance / Maximin / Maximax
™ Information on Attribute Given
¾ Conjunctive method / Lexicographic method /
SMART / AHP / ELECTRE / TOPSIS / ANP
™ Information on Alternative Given
¾ LINMAP

Hwang and Yoon 1981


MADM Environment: Criteria, Goals,
Attributes, Decision matrix
Example 1. A fighter selection problem
Decision Matrix

────────────────────────────────────────
Attributes
─────────────────────────────────
Alternatives Maximum Ferry Maximum Acquisition Reliability Maneuverability
( Ai ) speed range payload cost
(Mach) (NM) (pound) ($ X106) (high-low) (high-low)
───────────────────────────────────────
A1 2.0 1,500 20,000 5.5 average very high
A2 2.5 2,700 18,000 6.5 low average
A3 1.8 2,000 21,000 4.5 high high
A4 2.2 1,800 20,000 5.0 average average
────────────────────────────────────────

Hwang and Yoon 1981


Example 2: A fighter selection problem
No preference information given -- Maximin
# Decision matrix with assigning numerical values to qualitative
attributes:
x1 x2 x3 x4 x5 x6
⎡ 2.0 1,500 20,000 5.5 5 9 ⎤ A1
D= ⎢ 2.5 2,700 18,000 6.5 3 5 ⎥ A2
⎢ 1.8 2,000 21,000 4.5 7 7 ⎥ A3
⎣ 2.2 1,800 20,000 5.0 5 5 ⎦ A4

# Normalized decision matrix: Minimum


⎡ 0.80 0.56 0.95 0.82 0.71 1.00 ⎤ A1 ⎡0.56⎤
D*= ⎢ 1.00 1.00 0.86 0.69 0.43 0.56 ⎥ A2 ⎢0.43⎥
⎢ 0.72 0.74 1.00 1.00 1.00 0.78 ⎥ A3 ⎢0.72⎥
⎣ 0.88 0.67 0.95 0.90 0.71 0.56 ⎦ A4 ⎣0.56⎦
Maximum 0.72 ⇒ A3
Procedure for SAW
( Simple Additive Weighting Method )
1. Assign important weights for attributes
2. Make numerical scaling of intra-attribute values
3. Multiply the scale rating by its weight for each
attribute for each alternative
4. Sum up the scores of all attributes for each
alternative
5. Compare the total score of all alternatives, and the
alternative with the highest score is the one
prescribed to the DM
Example 3: A fighter selection problem
Simple additive weighted method
# DM assigns the weights to the attributes:
x1 x2 x3 x4 x5 x6
W = [ 0.2 0.1 0.1 0.1 0.2 0.3 ]

# Numerical scaling of intra-attribute values:


⎡0.80 0.56 0.95 0.82 0.71 1.00 ⎤ A1
D*= ⎢1.00 1.00 0.86 0.69 0.43 0.56 ⎥ A2
⎢0.72 0.74 1.00 1.00 1.00 0.78 ⎥ A3
⎣0.88 0.67 0.95 0.90 0.71 0.56 ⎦ A4

# Sum up the values concerned with each attribute for alternatives:


A1= 0.835, A2= 0.709, A3= 0.852, A4= 0.738 .

# Compare the scores of alternatives:


Thus, A3= 0.852 is selected.
Multiple Attribute Decision Making
™ Purpose: ranking and selection
™ Explicit Representation: ease of use
Multiple Objective Decision Making

Multiple Objective Optimization


Characteristics of MODM
™ Quantifiable Objectives
™ Constraints
™ A Trade-off Process

Hwang and Masud 1979


Methods for MODM (Stage at which
information is needed)
™ No Articulation of Prefer. Inform. Given
¾ Global criterion
™ A Priori Articulation of Prefer. Inform. Given
¾ Utility function / Lexicographic method / Goal
programming
™ Progressive Articulation of Prefer. Inform.
Given (interactive methods)
¾ SWT method / STEM method / Displaced ideal
™ A Posterior Articulation of Prefer. Inform.
Given
¾ Parametric method
Hwang and Masud 1979
Example 3. A Trade-off Problem
between Exports and Imports
Problem formulation
Max f1 = 2 x1 - x2 (effect on the export trade - 1st objective )
Max f2 = x1 + 2 x2 (profits on the product - 2nd objective )
s.t.
3 x1 - 5 x2 ≤ 15 ( capacity )
3 x1 - x2 ≤ 21 ( management )
3 x1 + x2 ≤ 27 ( space )
3 x1 + 4 x2 ≤ 45 ( material )
x1 + 3 x2 ≤ 30 ( labor hours )
x 1 , x2 ≥ 0 ( non-negative )

Shih et al. 1996


Decision (Variable) Space
Example 3. A Trade-off Problem
The Method of Global Criteria- minimize distance
function

k ⎧ ⎡ fi (x*) - fi (x) ⎤ P ⎫1/P


Min ∑ ⎨ ⎢ ⎯⎯⎯⎯⎯⎯⎯ ⎥ ⎬
I =1 ⎩ ⎣ fi (x*) ⎦ ⎭
s.t.
gi (x) ≥ 0 , i = 1, 2, ..., m.

P = 1 , Manhattan distance (geometric longest distance)


P = 2 , Euclidean distance (geometric shortest distance)
P = ∞ ,Tchebycheff distance (numerical shortest distance)
When P increases, the distance decreases

The objective function f = (9.33, 17.67) and decision x = (7.27, 5.2) for P
=2
Lee and Shih 2001
Objective (Function) Space
Multiple Objective Decision Making
™ Purpose: design
™ Implicit Representation: special structure
Group Decision Making

Organizational Decision Making


Characteristics of GDM
™ Multiple Criteria
™ Conflicts among Criteria
™ Committee

Hwang and Lin 1987


Methods for Group Decision Making
™ Social Choice Theory
¾ Voting / Social choice function / Social welfare
function
™ Experts Judgment / Group Participation
¾ Brainstorming / Brainwriting / NGT / Surveys /
Delphi / ISM / Cognitive map / PERT / CPM
™ Game Theory
¾ Normal form / Extensive form / Characteristic
function form

Hwang and Lin 1987


Example 4. A Voting Problem
Condorcet’s Function
# 60 voters vote from three candidates a, b, and c in the following
manner:
23 votes: a P c P b
19 votes: b P c P a
16 votes: c P b P a
2 votes: c P a P b

# Preferential voting system -- simple majority decision rule


⎡23 + 2 = 25 votes for a P b
⎣19 + 16 = 35 votes for b P a ⇒ b is better than a
⎡23 votes for a P c
⎣19 + 16 + 2 = 37 votes for c P a ⇒ c is better than a
⎡19 votes for b P c
⎣23 + 16 + 2 = 41 votes for c P b ⇒ c is better than b
ÖcPbPa
Hwang and Lin 1987
Group Decision Making
™ Purpose: creative confrontation, polling of
experts/participant, systematic structuring,
simulation, implementing and controlling
™ Implicit or Explicit Representation: variety
Applications of MCDM
™ Multiple Attribute Decision Making
¾ Commodity selection / Personnel selection /
Facility location / Public facility selection
™ Multiple Objective Decision Making
¾ System design / Production & manufacturing /
Capital budgeting / Manpower planning / Network
Flows / Socioeconomic problems
™ Group Decision Making
¾ Political voting / Commodity selection / Project
evaluation / Wargamming

Ö MCDM is used everywhere around us !!


Decision Support

Decision Aid
Decision Support
™ A total solution for problem-solving (or
decision-making)
™ Combinations of multiple techniques
™ Support by computers and network possibly
Example 5a. A recruitment and selection
processes (HRM)

Shih et al. 2005


Example 5b. A recruitment and selection
processes (HRM)
Information Systems
™ Decision support system
™ Group decision support system
™ Negotiation support system

Holsapple et al. 1996


Main Concerns

My Opinions
Making A Better/Satisfactory Decision
™ DM’s Preference
¾ Judgment/Reasonable basis/Interactive process
¾ Behavior and Psychology
™ Multiple Criteria
¾ Multiple dimensions for the essence
™ Group Decision
¾ Multiple persons with their knowledge and
experience
™ Risk/Uncertainty Management
¾ Dynamic environment/Time
™ Different Analysis Tools
¾ Result comparisons/Consistence
¾ Statistical tests
Future Studies
™ Validation of the model (experience)
™ Verification of effectiveness of different
techniques (decision quality)
™ Control of risk and uncertainty (reality)
™ Support by computerized processes and
techniques (effectiveness and efficiency)
™ Enhancement by soft computing (artificial
intelligence and bio-techniques)
References (I)
References (II)
References (III)
Conclusion

利之所在 弊即寓焉

福之所起 禍即伏焉
Questions & Comments

Thank you!

You might also like