Chapter 3 Structuring Decision 1 Structuring Decisions Learning

Chapter 3 • Structuring Decision 1

Structuring Decisions Learning Objectives • Fundamental steps in model creation • Identify and structure values and objectives – Fundamental objectives and hierarchies – Means objectives and networks • Graphical methods for decision frameworks – Influence diagrams – Decision trees • Concepts for model details – Elements – Probabilities – Cash flows – Objectives measurements 2

Structuring Decisions Three fundamental steps to create a decision model: 1. Identify and structure the values and objectives 2. Structure the decision elements into a logical framework 3. Refine and precisely define all elements of the model 3

Identifying and Structuring Values and Objectives • Identify important issues consistent with values • Identify and define relevant objectives • Organize objectives – Fundamental objectives and hierarchies – Means objectives and networks • Ensure consistency with context 4

Fundamental Objectives • General and reflect values • Organized in hierarchies • Paste figure 3. 1 5

Means Objectives • Identify how to accomplish fundamental objectives • Organized in networks • Paste figure 3. 2 6

Decision Context Three criteria for consistency of values-based objectives and decision context: 1. Properly reflective of decision situation 2. Decision owner has authority to make decision 3. Feasible to conduct analysis within resources 7

Structure Elements into Framework Two graphical methods: • Influence diagrams • Decision trees 8

Influence Diagrams • Geometric representation of decision elements: – Rectangles: represent decisions – Ovals: represent chance events – Diamonds: represent payoffs – Round-cornered rectangles: represent intermediate consequences or mathematical calculations 9

Influence Diagrams • Geometric shapes are called nodes – Rectangles: decision nodes – Ovals: chance nodes – Diamonds: payoff nodes – Round-cornered rectangles: consequence or calculation nodes 10

Influence Diagrams • Nodes connected by arcs to represent relationships • Nodes are named – Predecessor: if at beginning of arc – Successor: if at termination of arc Copy figure 3. 7 11

Influence Diagrams • Arcs represent two types of relationships, defined by the successor node • Arcs can represent sequence or relevance – Sequence: successor node is decision node – Relevance: successor node is any non-decision node Paste figure 3. 8 12

Influence Diagrams • Basic influence diagrams – Basic risky decision: One decision and one uncertainty – Imperfect information: Imperfect information about an uncertainty influences payoff – Sequential decision: Result from one decision determines if another decision is to be made – Intermediate calculation: Compiles predecessors information 13

Influence Diagrams Basic Risky Decision • Copy the figure 3. 9 Is the potential gain from choice A worth the risk that must be taken? 14

Influence Diagrams Imperfect Information • Copy the figure 3. 10 Imperfect information about uncertain event received and decision made; uncertain event is then resolved. Both decision and event affect payoff 15

Influence Diagrams Sequential Decisions • Copy the figure 3. 12 Sequential decisions reveal time sequence 16

Influence Diagrams Intermediate Calculations • Copy the figure 3. 16 Calculation nodes emphasize diagram structure 17

Creating Influence Diagrams • • No single strategy for creation Identify decision context and objectives Create simple version, then add details Unique representation rare 18

Creating Influence Diagrams Three common mistakes • Confusion with flow charts • Misuse of sequence arcs • Inclusion of cycles 19

Decision Trees • More details; sequential and chronological flows • Representation of events: – Square: decisions to be made – Circles: chance events • Branches from squares represent alternatives available • Branches from circles represent the possible outcomes of chance events • Final consequences or payoffs at branch ends • Decision trees flow from left to right 20

Decision Trees • Copy the figure 3. 21 – Alternatives: mutually exclusive; collectively exhaustive; select only one – Outcomes: mutually exclusive; collectively exhaustive; only one can occur – Complete tree includes all possible decision paths, alternatives and outcomes 21

Decision Trees • Multiple objectives: – List payoffs at branch ends – Can be cumbersome and bulky • Basic decision tree forms: – – – Basic risky decision Double- risk dilemma Range-of-risk dilemma Imperfect information Sequential decisions 22

Decision Trees Basic Risky Decision • Copy figure 3. 24 23

Decision Trees Double-Risk Dilemma • Copy figure 3. 26 24

Decision Trees Range-of-Risk Dilemma • Copy figure 3. 27 25

Decision Trees Imperfect Information • Copy figure 3. 28 26

Decision Trees Sequential Decisions • Copy figure 3. 29 Alternatives at second decision do not change as a result of outcome A or B 27

Decision Trees vs. Influence Diagrams • Decision trees – Display more information – Can become cumbersome • Influence Diagrams – Graphical presentation relatively simple – Easier for some to understand 28

Decision Trees vs. Influence Diagrams • Decision trees and influence diagrams are complementary • Strategy for use: – Start with influence diagram to understand major elements – Convert to decision tree to document details 29

Decision Details Defining elements • Elements must be measurable • Element definitions must require no judgment or interpretation 30

Decision Details Probabilities and Cash flows • Chance events require probability assignments – Only one outcome can occur – Probability of an outcome must be between 0 and 1 – Probabilities at chance node must sum to 1. 00 • Cash flows specified on branches – Cash flows compiled at branch ends to show consequences – Net present values used to reflect timing effects 31

Decision Details Measuring fundamental objectives • Measurement is crucial • Measure lowest level objectives in hierarchy • Measurement scales identified by attributes – Natural attribute scales – Proxy (surrogate) attribute scales – Constructed attribute scales 32

Summary • • Fundamental steps of model structuring Identify and structure values and objectives Graphical methods for structuring models Concepts for model details 33