Lecture 7: Decision Analysis – Part 2

OSCM 471/571: Optimization and Decision Support Modeling for Business

Seokjun Youn
([email protected])

Eller College of Management

University of Arizona
 Agenda
Announcement
…

Today’s Plan
 Lecture 7: Decision Analysis
• Reading: OptDSM_Ch9_Decision Analysis_Reading.pdf
1. A Case Study: The Goferbroke Company Problem (Sec 9.1)
2. Decision Trees (Sec 9.3)
3. Sensitivity Analysis with Decision Trees (Sec 9.4)
4. Checking Whether to Obtain More Information (Sec 9.5)
5. Using New Information to Update the Probabilities (Sec 9.6)
6. Decision Tree to Analyze a Sequence of Decisions (Sec 9.7)
7. Sensitivity Analysis with a Sequence of Decisions (Sec 9.8)
8. Using Utilities to Better Reflect the Values of Payoffs (Sec 9.9)

Next Class
 Lecture 8: Introduction to Simulation Modeling
A Case Study:
The Goferbroke Company Problem

3
The Goferbroke Company Problem
• The Goferbroke Company develops oil wells in unproven territory.
• A consulting geologist has reported that there is a one-in-four chance of oil on
a particular tract of land.
• Drilling for oil on this tract would require an investment of about $100,000.
• If the tract contains oil, it is estimated that the net revenue generated would
be approximately $800,000.
• Another oil company has offered to purchase the tract of land for $90,000.

Question:
Should Goferbroke drill for oil or sell the tract?

4
Prospective Profits

Status of Land Oil Profit Dry Profit

Alternative

Drill for oil $700,000 −$100,000

Sell the land 90,000 90,000

Chance of status 1 in 4 3 in 4

5
Data for Goferbroke Problem
• Prior Probabilities
State of Nature Prior Probability
The tract of land contains oil 0.25
The tract of land is dry (no oil) 0.75

• Payoff Table (Profit in $Thousands)

State of Nature
Alternative Oil Dry
Drill for oil 700 –100
Sell the land 90 90
Prior probability 0.25 0.75

6
Bayes’ Decision Rule
Bayes’ decision rule directly uses the prior probabilities.

Procedure:
• For each decision alternative, calculate the weighted average of its payoff by
multiplying each payoff by the prior probability and summing these products.
This is the expected payoff (EP).
• Choose the decision alternative that has the largest expected payoff.

7
Decision Trees using TreePlan

8
Decision Tree for Goferbroke
Payoff

700
Oil (0.25)

Drill
Dry (0.75)
-100
A

Sell

90

9
Using TreePlan to Construct Decision Trees
TreePlan can be used to construct decision trees on a spreadsheet.
1. Choose New># Branches from the Tree Plan ribbon.
2. Specify the type of node (Decision or Event).
3. Label the branches and specify the value for each branch.
4. Select the node at the end of the Drill branch and choose Decision Tree > Change Node
5. Choose Event node type and enter the name and partial payoffs for each branch.

10
TreePlan Results
• The numbers inside each decision node indicate which branch should be chosen (assuming
the branches are numbered consecutively from top to bottom).
• The numbers to the right of each terminal node is the payoff if that node is reached.
• The number 100 in cells A10 and E6 is the expected payoff at those stages in the process.

11
Sensitivity Analysis with Decision Trees

12
Consolidate the Data and Results

13
 Sensitivity Analysis
Prior Probability of Oil = 0.15 Prior Probability of Oil = 0.35

14
Using Data Tables to Do Sensitivity Analysis

15
Agenda
• Announcement
• Quiz 5: Apr 11, Please Study:
• Lecture Note 7: Decision Analysis Part 1 & 2
• Lecture Note 8: Introduction to Simulation Part 1
• Homework 5, Due: Apr 18, 11:59 pm

• Today’s Plan
• Lecture 7: Decision Analysis
• Reading: OptDSM_Ch9_Decision Analysis_Reading.pdf
1. Decision Tree to Analyze a Sequence of Decisions (Sec 9.7)
2. Sensitivity Analysis with a Sequence of Decisions (Sec 9.8)
3. Using Utilities to Better Reflect the Values of Payoffs (Sec 9.9)

• Next Class
• Lecture 8: Introduction to Simulation
Checking Whether to Obtain More Information

17
Checking Whether to Obtain More Information
• Might it be worthwhile to spend money for more information to obtain better
estimates?
• A quick way to check is to pretend that it is possible to actually determine the
true state of nature (“perfect information”).
• EP (with perfect information) = Expected payoff if the decision could be made
after learning the true state of nature.
• EP (without perfect information) = Expected payoff from applying Bayes’
decision rule with the original prior probabilities.
• The expected value of perfect information is then
EVPI = EP (with perfect information) − EP (without perfect information).

18
Expected Payoff with Perfect Information

By starting with an event node

involving the states of nature,
TreePlan uses this decision tree to
obtain the expected payoff with
perfect information for the first
Goferbroke Co. Problem.

EPwPI = 0.25*700+0.75*90 = 242.5

Previously,
EPwoPI was 100 (slide #13)

19
Using New Information to Update
the Probabilities in Spreadsheets

21
Using New Information to Update the Probabilities
• The prior probabilities of the possible states of nature often are quite
subjective in nature. They may only be rough estimates.

• It is frequently possible to do additional testing or surveying (at some

expense) to improve these estimates. The improved estimates are called
posterior probabilities.

22
Seismic Survey for Goferbroke
Goferbroke can obtain improved estimates of the chance of oil by conducting a
detailed seismic survey of the land, at a cost of $30,000.
Possible findings from a seismic survey:
• FSS: Favorable seismic soundings; oil is fairly likely.
• USS: Unfavorable seismic soundings; oil is quite unlikely.

P(finding|state) = Probability that the indicated finding will occur,

given that the state of nature is the indicated one.

23
 Calculating Joint Probabilities
• Each combination of a state of nature and a finding will have
a joint probability determined by the following formula:
P(state and finding) = P(state) P(finding|state)
• P(Oil and FSS) P=
= (Oil) P(FSS|Oil) = (0.25)(0.6) 0.15.

• P(Oil and USS) P=

= (Oil) P(USS|Oil) = (0.25)(0.4) 0.1.
• P(Dry and FSS) P=
= (Dry) P(FSS|Dry) = (0.75)(0.2) 0.15.

• P(Dry and USS) P=

= (Dry) P(USS|Dry) = (0.75)(0.8) 0.6.

24
Probabilities of Each Finding
• Given the joint probabilities of both a particular state of nature and a
particular finding, the next step is to use these probabilities to find each
probability of just a particular finding, without specifying the state of nature.

P(Finding) = P(Oil and Finding) + P(Dry and Finding)

• P(FSS) = 0.15 + 0.15 = 0.3.

• P(USS) = 0.1 + 0.6 = 0.7.

25
Calculating the Posterior Probabilities
• The posterior probabilities give the probability of a particular state of
nature, given a particular finding from the seismic survey.
P(state and finding)
P(state|finding) =
P(finding)
0.15
• P(Oil|FSS)
= = 0.5.
0.3
• P(Oil|USS) 0.1
= = 0.14.
0.7
0.15
• P(Dry|FSS)
= = 0.5.
0.3
0.6
• P(Dry|USS)
= = 0.86.
0.7
26
Probability Tree Diagram
Prior Conditional Joint Posterior
Probabilities Probabilities Probabilities Probabilities
P(state) P(finding | state) P(state and finding) P(state | finding)

0.15 = 0.5
0.25(0.6) = 0.15 0.3
0.6 Oil and FSS Oil, given FSS
FSS, given Oil

0.4
0.25 USS, given Oil 0.1 = 0.14
0.25(0.4) = 0.1 0.7
Oil
Oil and USS Oil, given USS

0.15 = 0.5
0.2 0.75(0.2) = 0.15 0.3
0.75 Dry, given FSS
Dry FSS, given Dry Dry and FSS

0.8 0.6 = 0.86

USS, given Dry 0.75(0.8) = 0.6 0.7
Dry and USS Dry, given USS

Unconditional probabilities: P(FSS) = 0.15 + 0.15 = 0.3

P(finding) P(USS) = 0.1 + 0.6 = 0.7
27
Posterior Probabilities

28
Template for Posterior Probabilities

29
Instant Quiz
A posterior probability is a revised probability of a state of nature after doing a test
or survey to refine the prior probability.
• True or False

Bayes' theorem is a formula for determining prior probabilities of a state of nature.

• True or False

30
Decision Tree to Analyze a
Sequence of Decisions

31
Decision Tree with Probabilities and Payoffs
Payoff
Oil (0.143) 670
800
f
Drill 0
-100 Dry(0.857) -130

c
90
Unfavorable Sell 60
0
Oil (0.5) 670
b 800
g
Do seismic survey 0 Drill 0
Favorable -100 Dry (0.5) -130
-30 (0.3)
d
90
a Sell 60

Oil (0.25) 700

800
0 h
Drill 0
No seismic survey -100 -100
Dry (0.75)
e
90
Sell 90 32
The Final Decision Tree
Payoff
Oil (0.143) 670
-15.7
800
f
0
Drill
-100 Dry (0.857) -130
60
c
90
Unfavorable Sell 60
0
123 Oil (0.5) 670
270
b g 800

0 Drill 0
Do seismic survey -100 Dry (0.5) -130
-30 Favorable (0.3) 270
d
123 90
a Sell 60
Oil (0.25) 700
100 800
0
h
No seismic survey Drill 0
-100 Dry (0.75) -100
100
e
90
Sell 90 33
TreePlan for the Full Goferbroke Co. Problem

34
Sensitivity Analysis with a
Sequence of Decisions

35
Organizing the Spreadsheet for Sensitivity Analysis

36
Data Table:
Optimal Policy vs. Prior Probability of Oil

Let p = Prior probability of oil

• If p ≤ 0.168, then sell the land (no seismic survey).
• If 0.169 ≤ p ≤ 0.308, then do the survey; drill if favorable, sell if not.
• If p ≥ 0.309, then drill for oil (no seismic survey). 37
Using Utilities to Better Reflect the
Values of Payoffs

38
Using Utilities to Better Reflect the Values of Payoffs
Thus far, when applying Bayes’ decision rule, we have assumed that the expected
payoff in monetary terms is the appropriate measure.
In many situations, this is inappropriate.
Suppose an individual is offered the following choice:
• Accept a 50-50 chance of winning $100,000.
• Receive $40,000 with certainty.

Many would pick $40,000, even though the expected payoff on the 50-50 chance of
winning $100,000 is $50,000. This is because of risk aversion.

A utility function for money is a way of transforming monetary values to an

appropriate scale that reflects a decision maker’s preferences (e.g., aversion to risk).
39
A Typical Utility Function for Money
U(M)

0.75

0.5

0.25

0
$10,000 $30,000 $60,000 $100,000 M
40
Shape of Utility Functions
U(M) U(M) U(M)

M M M
(a) Risk averse (b) Risk seeker (c) Risk neutral

• Risk averse curve slopes and bows upwards.

• Risk seeking curve slopes upward and bends downward.
• Risk neutral line is straight and upward sloping.
41
 Prospect theory
• Developed by Daniel Kahneman and
Amos Tversky in 1979.
• 2002 Nobel Memorial Prize in
Economics.
• Widely applied in behavioral
economics and behavioral finance.

The value function that passes through the

reference point is s-shaped and asymmetrical.
The value function is steeper for losses than
gains indicating that losses outweigh gains.

42
https://en.wikipedia.org/wiki/Prospect_theory
 Prospect theory

43
https://en.wikipedia.org/wiki/Prospect_theory
Utility Functions
• When a utility function for money is incorporated into a decision analysis
approach, it must be constructed to fit the current preferences and values of
the decision maker.

• Fundamental Property: Under the assumptions of utility theory, the decision

maker’s utility function for money has the property that the decision maker is
indifferent between two alternatives if the two alternatives have the same
expected utility.
• When the decision maker’s utility function for money is used, Bayes’ decision
rule replaces monetary payoffs by the corresponding utilities.
• The optimal decision (or series of decisions) is the one that maximizes the
expected utility.

44
Illustration of Fundamental Property
By the fundamental property, a decision maker with the utility function below-right will be
indifferent between each of the three pairs of alternatives below-left.

• 25% chance of $100,000. U(M)

• $10,000 for sure.

1
Both have E(Utility) = 0.25.

0.75
• 50% chance of $100,000.
• $30,000 for sure.
0.5
Both have E(Utility) = 0.5.

0.25
• 75% chance of $100,000.
• $60,000 for sure. 0
$10,000 $30,000 $60,000 $100,000 M
Both have E(Utility) = 0.75.
45
The Equivalent Lottery Method
1. Determine the largest potential payoff, M = Maximum.
Assign U(Maximum) = 1.
2. Determine the smallest potential payoff, M = Minimum.
Assign U(Minimum) = 0.
3. To determine the utility of another potential payoff M, consider the two
aleternatives:
A1: Obtain a payoff of Maximum with probability p.
Obtain a payoff of Minimum with probability 1 − p.
A2: Definitely obtain a payoff of M.

Question to the decision maker:

What value of p makes you indifferent?
Then, U(M) = p.

46
Generating the Utility Function for the
Decision Maker, Max Flyer
• The possible monetary payoffs in the Goferbroke Co. problem are −130, −100,
0, 60, 90, 670, and 700 (all in $thousands).
• Set U(Maximum) = U(700) = 1.
• Set U(Minimum) = U(−130) = 0.
• To find U(M), use the equivalent lottery method.
• For example, for M = 90, consider the two alternatives:
A1: Obtain a payoff of 700 with probability p
Obtain a payoff of −130 with probability 1 − p.
A2: Definitely obtain a payoff of 90.
1 1
=
• If Max chooses a point of indifference of p =, then U ( ) .
90
3 3

47
Max’s Utility Function for Money

48
Utilities for the Goferbroke Co. Problem

Monetary Payoff, M Utility, U(M)

–130 0.00

–100 0.05

60 0.30

90 0.33

670 0.97

700 1.00

49
Decision Tree with Max’s Utilities

50
Exponential Utility Function
• The procedure for constructing U(M) requires making many difficult decisions
about probabilities.
• An alternative approach assumes a certain form for the utility function and
adjusts this form to fit the decision maker as closely as possible.
• A popular form is the Exponential Utility Function
M
−
U ( M )= 1 − e R

where R is the decision maker’s risk tolerance.

• An easy way to estimate R is to pick the value that makes you indifferent between the
following two alternatives:
𝑅𝑅
a) A 50-50 gamble where you gain R dollars with probability 0.5 and lose dollars with
2

probability 0.5.

b) Neither gain nor lose anything.

51
Instant Quiz
Utilities can be useful when monetary values do not accurately reflect the true
values of an outcome.
• True or False

Two people who face the same problem and use the same decision-making
methodology must always arrive at the same decision.
• True or False

52
(Optional) What-If Analysis in Excel

• Data Tables
• https://www.excel-easy.com/examples/data-tables.html
• Scenario Manager
• https://www.excel-easy.com/data-analysis/what-if-analysis.html#create-
different-scenarios
• Goal Seek
• https://www.excel-easy.com/examples/goal-seek.html

53
What-If Analysis: Data Tables
• Instead of creating different scenarios, you can create a data table to
quickly try out different values for formulas.
• You can create a one variable data table or a two variable data table.

54
What-If Analysis: Scenario Manager
• What-If Analysis allows you to try out different values (scenarios) for formulas.

Result:

55
What-If Analysis: Goal Seek
• If you know the result you want from a formula, use Goal Seek in Excel
to find the input value that produces this formula result.
• E.g., Use Goal Seek in Excel to find the loan amount that produces a
monthly payment of $1500.

56