2-1
3 Fundamental Steps for Creating a Decision Model
Structuring Decisions (Modeling Decisions)
Identify and structure the values and
objectives
Structure the elements of the decision
situation into a logical framework
Refine and precisely define all of the
elements of the decision model
Dr. Yeu-Shiang Huang Managerial Decision Analysis
2-2
Modeling Process: Step 1
Structuring Decisions (Modeling Decisions)
Identify and Structure Values and
Objectives
Identify issues that matter to the decision
maker
List objectives and values and then separate
them into:
Fundamental objectives (hierarchy)
Means objectives (network)
Specify how to measure the accomplishment
of the objectives
Dr. Yeu-Shiang Huang Managerial Decision Analysis
1
2-3
Modeling Process: Step 2
Structuring Decisions (Modeling Decisions)
Structure the elements of the decision
situation into a logical framework
Two main tools for this step:
Influence diagrams
Decision trees
These two approaches are complementary
and have different pros and cons
Used in conjunction, they make a complete
model of the decision which show all of the
decision elements: Relevant objectives,
Decisions to be made, Uncertainties and
probabilities, Consequences
Dr. Yeu-Shiang Huang Managerial Decision Analysis
2-4
Modeling Process: Step 3
Structuring Decisions (Modeling Decisions)
Refine and precisely define all of the elements
of the decision model
Clearly and precisely understand
Decisions to be made
Available alternatives
Uncertain events involved
Measurement of consequences in terms of the
specified objectives
Dr. Yeu-Shiang Huang Managerial Decision Analysis
2
2-5
Modeling Process: Step 3
Structuring Decisions (Modeling Decisions)
Measuring the consequences and achievement of
objectives is especially important
Critical in allowing the comparison of the various
alternative
Two basic approaches to measure
Using natural scales when available
e.g., monetary measurement for items that can be
bought
Creating formal scales for non-quantitative
objectives
e.g., subjective scales for pleasure, fun, satisfaction,
etc.
Dr. Yeu-Shiang Huang Managerial Decision Analysis
2-6
Structuring Values
Structuring Decisions (Modeling Decisions)
Single objective vs. Multiple objectives
Understand what the objectives are
List all objectives and describe them in
detail
Balance multiple conflicting objectives
Determine direct and indirect effects
Direct are usually more important to
the decision
Dr. Yeu-Shiang Huang Managerial Decision Analysis
3
2-7
Structuring Values
Structuring Decisions (Modeling Decisions)
Techniques for identifying objectives
Develop a wish list
Identify alternatives
Consider problems and shortcomings
Predict consequences
Identify goals, constrains, and guidelines
Consider different perspectives
Determine strategic objectives
Determine generic objectives
Dr. Yeu-Shiang Huang Managerial Decision Analysis
2-8
Structuring Values
Structuring Decisions (Modeling Decisions)
Structuring the objectives
Objectives can be sorted into several
categories
Create categories that reflect the overall
objectives
Make sure the objectives are appropriate
for the decision context
Dr. Yeu-Shiang Huang Managerial Decision Analysis
4
2-9
Fundamental and Means Objectives
Structuring Decisions (Modeling Decisions)
Separate means from fundamental objectives
Fundamental objectives
Reflect what we really want to achieve
Are organized into hierarchies (FOH)
Upper level represent more general objectives
Lower level explain important elements of the
more general levels
Means objectives
Help achieve other objectives
Are organized into networks (MON)
Dr. Yeu-Shiang Huang Managerial Decision Analysis
2-10
Fundamental and Means Objectives
Structuring Decisions (Modeling Decisions)
Fundamental Objectives Hierarchies (FOH)
Dr. Yeu-Shiang Huang Managerial Decision Analysis
5
2-11
Fundamental and Means Objectives
Structuring Decisions (Modeling Decisions)
Means Objectives Networks (MON)
Dr. Yeu-Shiang Huang Managerial Decision Analysis
2-12
Fundamental and Means Objectives
Structuring Decisions (Modeling Decisions)
Structuring the FOH is crucial for developing a
multiple-objective decision model
Appropriate objectives (fundamental, NOT means )
are specified in the decision model
MON can suggest creative new alternatives
How to construct MON and FOH
Fundamental Means
Objectives Objectives
To Move: Downward in the Hierarchy: Away from Fundamental
Objectives:
Ask “What do you mean by that?” “How could you achieve this?”
To Move: Upward in the Hierarchy: Toward Fundamental Objectives:
“Of what more general objective “Why is that important?”
Ask is this an aspect?” (WITI)
Dr. Yeu-Shiang Huang Managerial Decision Analysis
6
2-13
Fundamental and Means Objectives
Structuring Decisions (Modeling Decisions)
The lowest-level fundamental objectives will
be the basis on which various consequences
will be measured
An easily measured means objective can
sometimes substitute for a fundamental
objective that is more difficult to measure
Networks and hierarchies provide insight
(e.g., what you care about, what criteria to
use) and suggest creative alternatives
Dr. Yeu-Shiang Huang Managerial Decision Analysis
2-14
Fundamental and Means Objectives
Structuring Decisions (Modeling Decisions)
MON and FOH
Dr. Yeu-Shiang Huang Managerial Decision Analysis
7
2-15
Getting the Decision Context Right
Structuring Decisions (Modeling Decisions)
Criteria for choosing a decision context
Scope: Are you addressing the right problem? Does
the context really capture the problem at hand?
Decision ownership: Do you have the authority to
make decisions within the specified context?
Feasibility: Will you be able to do the needed
analysis with the time and resources available?
Refining the context and modifying the objectives
are necessary
Decision situation is being taken seriously
Many different possibilities and perspectives are
being considered
Dr. Yeu-Shiang Huang Managerial Decision Analysis
2-16
Influence Diagrams
Structuring Decisions (Modeling Decisions)
Provide simple graphical representations of
decision situations
Components of an influence diagram
Rectangles: decisions
Ovals: chance events
Rounded rectangles: mathematical calculations
or constant values (intermediate consequences)
Diamonds: final consequence or payoff
Nodes and arrows (arcs)
Predecessors vs. Successors
Dr. Yeu-Shiang Huang Managerial Decision Analysis
8
2-17
Influence Diagrams
Structuring Decisions (Modeling Decisions)
Example Venture Succeeds
or Fails
Return on
Invest? Investment
Consequence is determined as soon as the
decision is made and the uncertain event is
resolved
No arc between the chance node and the
decision node
Dr. Yeu-Shiang Huang Managerial Decision Analysis
2-18
Influence Diagrams and the FOH
Structuring Decisions (Modeling Decisions)
FOH Maximize Overall Satisfaction
Invest in Computer Industry Return on Investment
Influence diagram
Venture Succeeds
Invest? or Fails
Return on Computer
Investment Industry Growth
Overall
Satisfaction
Dr. Yeu-Shiang Huang Managerial Decision Analysis
9
2-19
Arcs in Influence Diagrams
Structuring Decisions (Modeling Decisions)
Using arcs to represent relationships
An arc can represent
Relevance: when an arrow points into a chance or
consequence node
A B D E
C
F
Sequence: when an arrow points into a decision
node H
G
I
Properly constructed influence diagrams have NO
cycles
Dr. Yeu-Shiang Huang Managerial Decision Analysis
2-20
Common Mistakes for Influence Diagrams
Structuring Decisions (Modeling Decisions)
Interpreting influence diagrams as
flowcharts
an influence diagrams is a snapshot of the
decision situation at a particular point in time
Building influence diagrams with many
chance nodes having arrows pointing to
the primary decision node
Including cycles, i.e., circular paths
among the nodes
Dr. Yeu-Shiang Huang Managerial Decision Analysis
10
2-21
Decision Trees
Structuring Decisions (Modeling Decisions)
Display more of the details about a decision
Components of a decision tree
Squares: decisions
Circles: chance events
Branches
From a square: the choices available to the
decision maker
From a circle: the possible outcomes of a
chance event
Consequences (at the ends of the branches)
Dr. Yeu-Shiang Huang Managerial Decision Analysis
2-22
Decision Trees
Structuring Decisions (Modeling Decisions)
Example
Venture Succeeds Large return on
Investment
Invest
Venture Fails
Funds Lost
Do Not Invest "Typical" Return
Earned on Less
Risky Investment
Only one option allowed for each decision node
Outcomes from each chance node should be
mutually exclusive and collectively exhaustive
The decision tree shows all of the possible paths
Think of the nodes as occurring in a time sequence
Dr. Yeu-Shiang Huang Managerial Decision Analysis
11
2-23
Decision Trees and the FOH
Structuring Decisions (Modeling Decisions)
FOH
Safe and No Evacuation Cost Incurred
Maximize the Safety Minimize the Costs
Decision tree Storm Hits Miami
Safety Cost
Danger Low
Stay
Storm Misses Miami Safe Low
Forecast
Evacuate
Safe High
Dr. Yeu-Shiang Huang Managerial Decision Analysis
2-24
Decision Trees and Influence Diagrams
Structuring Decisions (Modeling Decisions)
Decision trees display considerably more
information
Decision trees get "messy" much faster as decision
problems become more complicated
Both are worthwhile (isomorphic), and complement
each other well
Influence diagrams are valuable for
The structuring phase of problem solving
Representing large problems
Decision trees are valuable for
Display the details of a problem
Dr. Yeu-Shiang Huang Managerial Decision Analysis
12
2-25
Decision Details: Elements of the Decision
Structuring Decisions (Modeling Decisions)
Make sure all elements of the decision
model are clearly defined
Different people have different idea
regarding some aspect of the decisions
Cash flows and probabilities
Clarity test: given perfect information,
could one tell exactly what happened at
every node, without judgment
Dr. Yeu-Shiang Huang Managerial Decision Analysis
2-26
Decision Details: Elements of the Decision
Structuring Decisions (Modeling Decisions)
Defining measurement scales for
fundamental objectives
Unambiguous
Attribute scale (attribute)
Natural scale
Artificial scale
Proxy – find a substitute for an attribute
Levels – construct a scale of levels for an
attribute that can then facilitate
measurement
Dr. Yeu-Shiang Huang Managerial Decision Analysis
13