Click to edit Master title style

Decision Analysis
D r. M u k e s h K u m a r M e h l a wa t

1
Click to edit Master title style

Introduction

2 2
Click to
What is decision
edit Master
analysis
title style

In integer programming, we focused mainly on decision making when the consequences of

alternative decisions are known with a reasonable degree of certainty.
This decision-making environment enabled formulating helpful mathematical models (linear
programming, integer programming, nonlinear programming, etc.) with objective functions that
specify the estimated consequences of any combination of decisions.
Although these consequences usually cannot be predicted with complete certainty, they could
at least be estimated with enough accuracy to justify using such models
However, decisions often must be made in environments that are much more fraught with
uncertainty.
Decision analysis provides a framework and methodology for rational decision making when
the outcomes are uncertain.

3 3
Click to
What is decision
edit Master
analysis
title style

Here are a few examples:

• A manufacturer introducing a new product into the marketplace. What will be the reaction of
potential customers? How much should be produced? Should the product be test marketed
in a small region before deciding upon full distribution? How much advertising is needed to
launch the product successfully?
• A financial firm investing in securities. Which are the market sectors and individual securities
with the best prospects? Where is the economy headed? How about interest rates? How
should these factors affect the investment decisions?
• A government contractor bidding on a new contract. What will be the actual costs of the
project? Which other companies might be bidding? What are their likely bids?

4 4
Click to
What is decision
edit Master
analysis
title style
• An agricultural firm selecting the mix of crops and livestock for the upcoming season. What
will be the weather conditions? Where are prices headed? What will costs be?
• An oil company deciding whether to drill for oil in a particular location. How likely is oil there?
How much? How deep will they need to drill? Should geologists investigate the site further
before drilling?
These are the kinds of decision making in the face of great uncertainty that decision analysis is
designed to address.
Frequently, one question to be addressed with decision analysis is whether to make the
needed decision immediately or to first do some testing (at some expense) to reduce the level
of uncertainty about the outcome of the decision.
For example, the testing might be field testing of a proposed new product to test consumer
reaction before making a decision on whether to proceed with full-scale production and
marketing of the product. This testing is referred to as performing experimentation. Therefore,
decision analysis divides decision making between the cases of without experimentation and
with experimentation.
5 5
 Decision
Click to editAnalysis
Master title style

• Decision Making under Uncertainty

• Several (finite number) of Decision Alternatives
• Outcome of the decision depends on a Chance Event which is yet to
take place (State of Nature)
• State of Nature
• A chance event that is relevant to the decision, but beyond our control
• The event may have several (finite) possible outcomes.
• Actual outcome will be known only after the decision has been taken.
• Probability distribution of different states of nature may or may not be known.

6
Click to editAnalysis
Decision Master title style

• How to decide which decision alternative to select?

• Example
• Cricket Match in Delhi
• May get cancelled in case of rain
• Should you go or not?
• Decision Alternatives: Go Don’t Go
• States of Nature Rain No Rain

7
Click to edit
Payoff Master title style
Table

• There is a payoff for each combination of decision chosen, and

the state of nature.
• Express the payoff in the form of a table.

Rain No Rain
Go
-100 500

Don’t Go
0 0

8
Click to editExample
Another Master title style

• An oil company owns a land, that may contain oil

• Decision Alternative: Drill for oil Sell
• States of Nature: Oil Dry

Oil Dry
Drill 700,000 -100,000

Sell 90,000 90,000

9
Click
Howto to
edit Master title style
Decide?

• When probabilities of states of nature are not available.

• Several possible decision criteria
• No universal right decision
• Depends on the outlook of the decision maker
• Pessimistic, Optimistic, partly Optimistic.
• MaxiMax Payoff (Optimistic)
• MaxiMin Payoff (Pessimistic)
• Hurvicz’s Criterion (partly Optimistic)
• MiniMax Regret Criterion

10
Click to editPayoff
MaxiMax Master title style

• Pick the decision alternative corresponding to the maximum

payoff figure in the entire Payoff Table.
Rain No Rain
Go
-100
-100,000 500

Don’t Go
0 0

Drawback: Ignores all entries in the Payoff Tables except one.

Can lead to a big loss if the state of nature is not favorable.

11
 MaxiMin
Click to editPayoff
Master title style

• Find the worst case payoff for each decision alternative

• Select the decision that maximizes the worst case payoff
Rain No Rain Min
Go
-100 500,000
500 -100

Don’t Go
0 0 0

Max = 0

Drawback: Too pessimistic. Assumes that the worst will happen.

Ignores payoff values other than minimum in the row.
12
Click to edit Master
Hurwicz’s title style
Criterion

• Lets you choose a level of optimism α between 0 and 1

• For each decision alternative, Compute Weighted Payoff
= α(Maximum) + (1-α ).(Minimum)
• Select Decision Alternative with Maximum Weighted Payoff.
α=0.4 Rain No Rain Wt. Payoff
Go
-100 500 500*0.4-100*0.6 140

Don’t Go
0 0 0

13
Click to editRegret
MiniMax MasterCriterion
title style

• Compute a Regret Table

• For a given combination of decision alternative and state of nature
• Regret = Best possible payoff for that state of nature – Actual payoff
• Choose decision that minimizes the maximum regret
Regret Table
Rain No Rain Rain No Rain Max
Go 0-(-100) 500-(500)
-100 500 100
100 0
Don’t Go
0 0 0 500 500

14
 Example
Click to edit Master title
S1 style S2 S3
D1 30 5 -10
D2 40 10 -30
D3 -12 0 15

S1 S2 S3
D1 10 5 25 25
D2 0 0 45 45
52
D3 52 10 0

15
Click to edit Master
Decision-making under
titleuncertainty
style
National Outdoors School (NOS) is preparing a summer campsite in the heart of Alaska to train
individuals in wilderness survival. NOS estimates that attendance can fall into one of four
categories: 200, 250, 300, and 350 persons. The cost of the campsite will be the smallest
when its size meets the demand exactly. Deviations above or below the ideal demand levels
incur additional costs resulting from constructing more capacity than needed or losing income
opportunities when the demand is not met. Letting 𝑎𝑎1 to 𝑎𝑎4 represent the sizes of the
campsites (200, 250, 300, and 350 persons) and 𝑠𝑠1 to 𝑠𝑠4 the level of attendance, the following
table summarizes the cost matrix (in thousands of dollars) for the situation,

𝑠𝑠1 𝑠𝑠2 𝑠𝑠3 𝑠𝑠4

𝑎𝑎1 5 8 18 25
𝑎𝑎2 8 7 12 23
𝑎𝑎3 21 18 12 21
𝑎𝑎4 30 22 19 15

The problem is analysed using all four criteria.

1616
Click
A prototype
to edit example
Master title style
The GOFERBROKE COMPANY owns a tract of land that may contain oil. A consulting
geologist has reported to management that she believes there is 1 chance in 4 of oil.
Because of this prospect, another oil company has offered to purchase the land for $90,000.
However, Goferbroke is considering holding the land in order to drill for oil itself. The cost of
drilling is $100,000. If oil is found, the resulting expected revenue will be $800,000, so the
company’s expected profit (after deducting the cost of drilling) will be $700,000. A loss of
$100,000 (the drilling cost) will be incurred if the land is dry (no oil).
The following table (next slide) summarises this information.
The next section discusses how to approach the decision of whether to drill or sell based just
on these data. (We will refer to this as the first Goferbroke Co. problem.)
However, before deciding whether to drill or sell, another option is to conduct a detailed
seismic survey of the land to obtain a better estimate of the probability of finding oil. The next
section also discusses this case of decision making with experimentation, at which point the
necessary additional data will be provided.

1717
Click
A prototype
to edit example
Master title style
Payoff
Alternative
Oil Dry

Drill for oil $ 700,000 -$ 100,000

Sell the land $ 90,000 $ 90,000

Chance of
1 in 4 3 in 4
status

This company is operating without much capital, so a loss of $100,000 would be quite serious.
In the next section, we describe how to refine the evaluation of the consequences of the
various possible outcomes.

1818
Click to edit Master
Decision-making without
title style
experimentation
As indicated in the table of the Goferbroke Co. has two possible actions under consideration:
drill for oil or sell the land. The possible states of nature are that the land contains oil and that it
does not, as designated in the column headings of the table by oil and dry. Since the
consulting geologist has estimated that there is 1 chance in 4 of oil (and so 3 chances in 4 of
no oil), the prior probabilities of the two states of nature are 0.25 and 0.75, respectively. We
present the data here again in the payoff matrix format (units in 1000s of dollars).

State of nature
Alternatives
Oil Dry

Drill the land 700 -100

Sell the land 90 90

Prior probability 0.25 0.75

1919
Click to edit Master
Decision-making without
title style
experimentation
The rationale for this criterion is that it provides the best guarantee of the payoff that will be
obtained. Regardless of what the true state of nature turns out to be for the example, the
payoff from selling the land cannot be less than 90, which provides the best available
guarantee. Thus, this criterion takes the pessimistic viewpoint that, regardless of which action
is selected, the worst state of nature for that action is likely to occur, so we should choose the
action which provides the best payoff with its worst state of nature.
This rationale is quite valid when one is competing against a rational and malevolent opponent.
However, this criterion is not often used in games against nature because it is an extremely
conservative criterion in this context. In effect, it assumes that nature is a conscious opponent
that wants to inflict as much damage as possible on the decision maker.
Nature is not a malevolent opponent, and the decision maker does not need to focus solely on
the worst possible payoff from each action. This is especially true when the worst possible
payoff from an action comes from a relatively unlikely state of nature.

2020
Click to edit Master
Decision-making without
title style
experimentation
The Maximum likelihood criterion: Identify the most likely state of nature (the one with the
largest prior probability). For this state of nature, find the action with the maximum payoff.
Choose this action.
Applying this criterion to the example, the table indicates that the Dry state has the largest prior
probability. In the Dry column, the sell alternative has the maximum payoff, so the choice is to
sell the land.

2121
Click to edit Master
Decision-making without
title style
experimentation
The appeal of this criterion is that the most important state of nature is the most likely one, so
the action chosen is the best one for this particularly important state of nature.
Basing the decision on the assumption that this state of nature will occur tends to give a better
chance of a favourable outcome than assuming any other state of nature. Furthermore, the
criterion does not rely on questionable subjective estimates of the probabilities of the
respective states of nature other than identifying the most likely state.
The major drawback of the criterion is that it completely ignores much relevant information. No
state of nature is considered other than the most likely one. In a problem with many possible
states of nature, the probability of the most likely one may be quite small, so focusing on just
this one state of nature is quite unwarranted.
Even in the example, where the prior probability of the Dry state is 0.75, this criterion ignores
the extremely attractive payoff of 700 if the company drills and finds oil. In effect, the criterion
does not permit gambling on a low-probability big payoff, no matter how attractive the gamble
may be.

2222
Click to edit Master
Decision-making without
title style
experimentation
Bayes’ Decision rule: Using the best available estimates of the probabilities of the respective
states of nature (currently the prior probabilities), calculate the expected value of the payoff for
each of the possible actions. Choose the action with the maximum expected payoff.
For the prototype example, these expected payoffs are calculated directly as,
𝐸𝐸 Payoff drill = 0.25 700 + 0.75 −100
= 100.
𝐸𝐸[Payoff (sell)] = 0.25(90) + 0.75(90)
= 90.
Since 100 is larger than 90, the alternative action selected is to drill for oil. Note that this choice
contrasts with the selection of the sell alternative under each of the two preceding criteria.

2323
Click to edit Master
Decision-making without
title style
experimentation
The big advantage of Bayes’ decision rule is that it incorporates all the available information,
including all the payoffs and the best available estimates of the probabilities of the respective
states of nature.
It is sometimes argued that these estimates of the probabilities necessarily are largely
subjective and so are too shaky to be trusted. There is no accurate way of predicting the
future, including a future state of nature, even in probability terms. This argument has some
validity. The reasonableness of the estimates of the probabilities should be assessed in each
individual situation.
Nevertheless, under many circumstances, past experience and current evidence enable one to
develop reasonable estimates of the probabilities. Using this information should provide better
grounds for a sound decision than ignoring it. Furthermore, experimentation frequently can be
conducted to improve these estimates, as described in the next section. Therefore, we will be
using only Bayes’ decision rule throughout the remainder of the presentation.
To assess the effect of possible inaccuracies in the prior probabilities, it often is helpful to
conduct sensitivity analysis, as described next.
2424
Click to edit Master
Decision-making without
title style
experimentation
Sensitivity Analysis with Bayes’ Decision Rule
Sensitivity analysis commonly is used with various applications of operations research to study
the effect if some of the numbers included in the mathematical model are not correct. In this
case, the mathematical model is represented by the payoff table.
The numbers in this table that are most questionable are the prior probabilities. We will focus
the sensitivity analysis on these numbers, although a similar approach could be applied to the
payoffs given in the table.
The sum of the two prior probabilities must equal 1, so increasing one of these probabilities
automatically decreases the other one by the same amount, and vice versa. Goferbroke’s
management feels that the true chances of having oil on the tract of land are likely to lie
somewhere between 15 and 35 percent.
Sensitivity analysis begins by reapplying Bayes’ decision rule twice, once when the prior
probability of oil is at the lower end of this range (0.15) and next when it is at the upper end
(0.35).

2525
Click to edit Master
Decision-making without
title style
experimentation

2626
Click to edit Master
Decision-making without
title style
experimentation
When the prior probability of oil is only 0.15, the decision swings over to selling the land by a
wide margin (an expected payoff of 90 versus only 20 for drilling). However, when this
probability is 0.35, the decision is to drill by a wide margin (expected payoff ? 180 versus only
90 for selling). Thus, the decision is very sensitive to the prior probability of oil. This sensitivity
analysis has revealed that it is important to do more, if possible, to pin down just what the true
value of the probability of oil is.
Let 𝑝𝑝 be the prior probability of oil.
The expected payoff from drilling for any 𝑝𝑝 is,
𝐸𝐸 Payoff (drill) = 700𝑝𝑝 + 100(1 − 𝑝𝑝)
= 800𝑝𝑝 + 100.

2727
Click to edit Master
Decision-making without
title style
experimentation

2828
Click to edit Master
Decision-making without
title style
experimentation
The slanting line in the graph shows the plot of this expected payoff versus 𝑝𝑝, which is just the
line passing through the two points given by cells C10 and H5 in the two spreadsheets.
The point in the graph where the two lines intersect is the crossover point where the decision
shifts from one alternative (sell the land) to the other (drill for oil) as the prior probability
increases. To find this point, we set,
𝐸𝐸 Payoff drill = 𝐸𝐸[Payoff (sell)]
800𝑝𝑝 − 100 = 90
190
𝑝𝑝 = = 0.2375.
800
Conclusion: Should sell the land if 𝑝𝑝 < 0.2375, should drill for oil if 𝑝𝑝 > 0.2375.
For other problems that have more than two alternative actions, the same kind of analysis can
be applied. The main difference is that there now would be more than two lines in the graphical
display. However, the top line for any particular value of the prior probability still indicates which
alternative should be chosen. With more than two lines, there might be more than one
crossover point where the decision shifts from one alternative to another.
2929
Click to edit Master
Decision-making without
title style
experimentation
For a problem with more than two possible states of nature, the most straightforward approach
is to focus the sensitivity analysis on only two states at a time as described above. This again
would involve investigating what happens when the prior probability of one state increases as
the prior probability of the other state decreases by the same amount, holding fixed the prior
probabilities of the remaining states. This procedure then can be repeated for as many other
pairs of states as desired.

Because the decision the Goferbroke Co. should make depends so critically on the true
probability of oil, serious consideration should be given to conducting a seismic survey to
estimate this probability more closely. We will explore this option in the next two sections.

3030
Click to edit Master
Decision-making with
title
experimentation
style
Frequently, additional testing (experimentation) can be done to improve the preliminary
estimates of the probabilities of the respective states of nature provided by the prior
probabilities. These improved estimates are called posterior probabilities.
Continuing the Prototype Example, an available option before making a decision is to
conduct a detailed seismic survey of the land to obtain a better estimate of the probability of oil.
The cost is $30,000. A seismic survey obtains seismic soundings that indicate whether the
geological structure is favourable to the presence of oil. We will divide the possible findings of
the survey into the following two categories:
USS: Unfavourable seismic soundings; oil is fairly unlikely.
FSS: Favourable seismic soundings; oil is fairly likely.
Based on past experience, if there is oil, then the probability of unfavorable seismic sound-
ings is
𝑃𝑃 USS State=Oil) = 0.4, so 𝑃𝑃 FSS|State =Oil = 1 − 0.4 = 0.6.

3131
Click to edit Master
Decision-making with
title
experimentation
style
Similarly, if there is no oil (i.e., the true state of nature is Dry), then the probability of
unfavourable seismic soundings is estimated to be
𝑃𝑃 USS|State=Dry = 0.8, so 𝑃𝑃 FSS|State =Dry = 1 − 0.8 = 0.2.
Posterior Probabilities
Proceeding now in general terms, we let,
𝑛𝑛 = number of possible states of nature;
𝑃𝑃 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆 = 𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 𝑖𝑖 = prior probability that true state of nature is state i, for 𝑖𝑖 = 1, 2, . . . , 𝑛𝑛;
Finding = finding from experimentation (a random variable);
Finding 𝑗𝑗 = one possible value of finding;
posterior probability that true state of nature is state 𝑖𝑖,
𝑃𝑃 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆 = 𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 𝑖𝑖 𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹 = 𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑔𝑔 𝑗𝑗 =
given that Finding = finding 𝑗𝑗, for 𝑖𝑖 = 1, 2, . . . , 𝑛𝑛.
The question currently being addressed is the following:
Given 𝑃𝑃(𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆 = 𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 𝑖𝑖) and 𝑃𝑃(𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹 = 𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓 𝑗𝑗|𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆 = 𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 𝑖𝑖), for 𝑖𝑖 = 1, 2, . . . , 𝑛𝑛, what is
𝑃𝑃 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆 = 𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 𝑖𝑖 𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹 = 𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓 𝑗𝑗 ?

3232
Click to edit Master
Decision-making with
title
experimentation
style
This question is answered by combining the following standard formulas of probability theory:
𝑃𝑃 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆 = 𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 𝑖𝑖, 𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹 = 𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓 𝑗𝑗
𝑃𝑃 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆 = 𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 𝑖𝑖 𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹 = 𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓 𝑗𝑗 = .
𝑃𝑃 𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹 = 𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓 𝑗𝑗
𝑛𝑛

𝑃𝑃 𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹 = 𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓 𝑗𝑗 = � 𝑃𝑃 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆 = 𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 𝑘𝑘, 𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹 = 𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓 𝑗𝑗 .

𝑘𝑘=1
𝑃𝑃 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆 = 𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 𝑖𝑖, 𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹 = 𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓 𝑗𝑗 = 𝑃𝑃 𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹 = 𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓 𝑗𝑗 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆 = 𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 𝑖𝑖 𝑃𝑃 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆 = 𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 𝑖𝑖 .

Therefore, for each 𝑖𝑖 = 1, 2, . . . , 𝑛𝑛, the desired formula for the corresponding posterior
probability is
𝑃𝑃 𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹 = 𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓 𝑗𝑗 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆 = 𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 𝑖𝑖 𝑃𝑃 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆 = 𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 𝑖𝑖
𝑃𝑃 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆 = 𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 𝑖𝑖 𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹 = 𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓 𝑗𝑗 = .
∑𝑛𝑛𝑘𝑘=1 𝑃𝑃 𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹 = 𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓 𝑗𝑗 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆 = 𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 𝑘𝑘 𝑃𝑃 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆 = 𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 𝑘𝑘

3333
Click to edit Master
Decision-making with
title
experimentation
style
Now let us return to the prototype example and apply this formula. If the finding of the
seismic survey is unfavourable seismic soundings (USS), then the posterior probabilities
are,
0.4(0.25) 1
𝑃𝑃 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆 = 𝑂𝑂𝑂𝑂𝑂𝑂 𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹 = 𝑈𝑈𝑈𝑈𝑈𝑈 = = .
0.4 0.25 + 0.8(0.75) 7
1 6
𝑃𝑃 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆 = 𝐷𝐷𝐷𝐷𝐷𝐷 𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹 = 𝑈𝑈𝑈𝑈𝑈𝑈 = 1 − = .
7 7
0.6(0.25) 1
𝑃𝑃 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆 = 𝑂𝑂𝑂𝑂𝑂𝑂 𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹 = 𝐹𝐹𝑆𝑆𝑆𝑆 = = .
0.6 0.25 + 0.2(0.75) 2
1 1
𝑃𝑃 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆 = 𝐷𝐷𝐷𝐷𝐷𝐷 𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹 = 𝐹𝐹𝐹𝐹𝐹𝐹 = 1 − = .
2 2
The probability tree diagram shows a nice way of organizing these calculations in an
intuitive manner. The prior probabilities in the first column and the conditional probabilities
in the second column are part of the input data for the problem. Multiplying each
probability in the first column by a probability in the second column gives the
corresponding joint probability in the third column. Each joint probability then becomes the
numerator in the calculation of the corresponding posterior probability in the fourth column.
3434
Click to edit Master
Decision-making with
title
experimentation
style

3535
Click to edit Master
Decision-making with
title
experimentation
style
After these computations have been completed, Bayes’ decision rule can be applied just as
before, with the posterior probabilities now replacing the prior probabilities. Again, by using the
payoffs (in units of thousands of dollars) from the table and subtracting the cost of the
experimentation, we obtain the results shown below.
Expected payoffs if finding is unfavourable seismic soundings (USS):
1 6
𝐸𝐸 Payoff drill |Finding=USS = 700 + −100 − 30
7 7
= −15.7.
1 6
𝐸𝐸[Payoff (sell)|Finding=USS] = 90 + 90 − 30
7 7
= 60.
Expected payoffs if finding is favourable seismic soundings (FSS):
1 1
𝐸𝐸 Payoff(drill)|Finding=FSS = 700 + −100 − 30
2 2
= 270.
1 1
𝐸𝐸 Payoff(sell)|Finding=FSS = 90 + 90 − 30 = 60.
2 2 3636
Click to edit Master
Decision-making with
title
experimentation
style
Since the objective is to maximize the expected payoff, these results yield the optimal policy
shown in the table.

Finding from the Optimal Action Expected payoff Expected payoff

seismic survey excluding the cost including cost of
of survey survey
USS Sell the land 90 60

FSS Drill for oil 300 270

However, what this analysis does not answer is whether it is worth spending $30,000 to
conduct the experimentation (the seismic survey). Perhaps it would be better to forgo this
major expense and just use the optimal solution without experimentation (drill for oil, with an
expected payoff of $100,000). We address this issue next.

3737
Click to edit Master
Decision-making with
title
experimentation
style
The value of Experimentation
The first method assumes (unrealistically) that the experiment will remove all uncertainty about
what the true state of nature is, and then this method makes a very quick calculation of what
the resulting improvement in the expected payoff would be (ignoring the cost of the
experiment). This quantity, called the expected value of perfect information, provides an
upper bound on the potential value of the experiment. Therefore, if this upper bound is less
than the cost of the experiment, the experiment definitely should be forgone.
Expected Value of Perfect Information: Suppose now that the experiment could definitely
identify what the true state of nature is, thereby providing “perfect” information. Whichever
state of nature is identified, you naturally choose the action with the maximum payoff for that
state. We do not know in advance which state of nature will be identified, so a calculation of
the expected payoff with perfect information (ignoring the cost of the experiment) requires
weighting the maximum payoff for each state of nature by the prior probability of that state of
nature.
We have,Expected payoff with perfect information=0.25 700 + 0.75 90 = 242.5.

3838
Click to edit Master
Decision-making with
title
experimentation
style

Thus, if the Goferbroke Co. could learn before choosing its action whether the land contains
oil, the expected payoff as of now (before acquiring this information) would be $242,500
(excluding the cost of the experiment generating the information.) To evaluate whether the
experiment should be conducted, we now use this quantity to calculate the expected value of
perfect information. The expected value of perfect information, abbreviated EVPI, is calculated
as,
𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸
= 𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒 𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝 𝑤𝑤𝑤𝑤𝑤𝑤𝑤 𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝 𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 − 𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒 𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝 𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤 𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒. 3939
Click to edit Master
Decision-making with
title
experimentation
style
For the prototype example, we found that the expected payoff without experimentation (under
Bayes’ decision rule) is 100. Therefore,
𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸 = 242.5 − 100 = 142.5.
Since 142.5 far exceeds 30, the cost of experimentation (a seismic survey), it may be
worthwhile to proceed with the seismic survey. To find out for sure, we now go to the second
method of evaluating the potential benefit of experimentation.
Expected Value of Experimentation: Rather than just obtain an upper bound on the expected
increase in payoff (excluding the cost of the experiment) due to performing experimentation,
we now will do somewhat more work to calculate this expected increase directly. This quantity
is called the expected value of experimentation.
Calculating this quantity requires first computing the expected payoff with experimentation
(excluding the cost of the experiment). Obtaining this latter quantity requires doing all the work
described earlier to find all the posterior probabilities, the resulting optimal policy with
experimentation, and the corresponding expected payoff (excluding the cost of the experiment)
for each possible finding from the experiment. Then each of these expected payoffs needs to
be weighted by the probability of the corresponding finding, that is,
Expected payoff with experimentation= ∑𝑗𝑗 𝑃𝑃 𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹 = 𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓 𝑗𝑗 𝐸𝐸[𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝|𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹 = 𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓 𝑗𝑗]. 4040
Click to edit Master
Decision-making with
title
experimentation
style
For the prototype example, we have already done all the work to obtain the terms on the right
side of this equation. The values of 𝑃𝑃(𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹 = 𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓 𝑗𝑗) for the two possible findings from
the seismic survey—unfavourable (USS) and favourable (FSS)—were calculated at the bottom
of the probability tree diagram as 𝑃𝑃 𝑈𝑈𝑈𝑈𝑈𝑈 = 0.7, 𝑃𝑃 𝐹𝐹𝐹𝐹𝐹𝐹 = 0.3. For the optimal policy with
experimentation, the corresponding expected payoff (excluding the cost of the seismic survey)
for each finding was obtained in the third column of the table as,
𝐸𝐸 𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃 𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹 = 𝑈𝑈𝑈𝑈𝑈𝑈 = 90.
𝐸𝐸 𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃 𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹 = 𝐹𝐹𝐹𝐹𝐹𝐹 = 300.
Therefore,
Expected payoff with experimentation= 0.7 90 + 0.3 300 = 153.
Expected value of experimentation= 153 − 100 = 53.
Since this value exceeds 30, the cost of conducting a detailed seismic survey (in units of
thousands of dollars), this experimentation should be done.

4141
Click to edit
Decision trees
Master title style
Decision trees provide a useful way of visually displaying the problem and then organising the
computational work already described in the preceding two situations. These trees are
especially helpful when a sequence of decisions must be made.
Constructing the Decision Tree
The prototype example involves a sequence of two decisions:
1. Should a seismic survey be conducted before an action is chosen?
2. Which action (drill for oil or sell the land) should be chosen?
The nodes of the decision tree are referred to as forks, and the arcs are called branches. A
decision fork, represented by a square, indicates that a decision needs to be made at that
point in the process. A chance fork, represented by a circle, indicates that a random event
occurs at that point.

4242
Click to edit
Decision trees
Master title style

4343
Click to edit
Decision trees
Master title style
Thus, in the figure, the first decision is represented by decision fork 𝑎𝑎. Fork b is a chance fork
representing the random event of the outcome of the seismic survey. The two branches
emanating from fork b represent the two possible outcomes of the survey. Next comes the
second decision (forks 𝑐𝑐, 𝑑𝑑, and 𝑒𝑒) with its two possible choices. If the decision is to drill for oil,
then we come to another chance fork (forks 𝑓𝑓, 𝑔𝑔, and ℎ), where its two branches correspond to
the two possible states of nature.
Note that the path followed from fork a to reach any terminal branch (except the bottom one) is
determined both by the decisions made and by random events that are outside the control of
the decision maker. This is characteristic of problems addressed by decision analysis.
The next step in constructing the decision tree is to insert numbers into the tree as shown in
the next figure. The numbers under or over the branches that are not in parentheses are the
cash flows (in thousands of dollars) that occur at those branches. For each path through the
tree from node 𝑎𝑎 to a terminal branch, these same numbers then are added to obtain the
resulting total payoff shown in boldface to the right of that branch. The last set of numbers is
the probabilities of random events. In particular, since each branch emanating from a chance
fork represents a possible random event, the probability of this event occurring from this fork
has been inserted in parentheses along this branch.
4444
Click to edit
Decision trees
Master title style

4545
Click to edit
Decision trees
Master title style
From chance fork ℎ, the probabilities are the prior probabilities of these states of nature,
since no seismic survey has been conducted to obtain more information in this case. However,
chance forks 𝑓𝑓 and 𝑔𝑔 lead out of a decision to do the seismic survey (and then to drill).
Therefore, the probabilities from these chance forks are the posterior probabilities of the
states of nature, given the finding from the seismic survey (the data that we calculated in the
previous slide. Finally, we have the two branches emanating from chance fork b. The numbers
here are the probabilities of these findings from the seismic survey, Favourable (FSS) or
Unfavourable (USS), as given underneath the probability tree diagram.
Performing the analysis
Having constructed the decision tree, including its numbers, we now are ready to analyze the
problem by using the following procedure.
1. Start at the right side of the decision tree and move left one column at a time. For each
column, perform either step 2 or step 3 depending upon whether the forks in that column
are chance forks or decision forks.

4646
Click to edit
Decision trees
Master title style
2. For each chance fork, calculate its expected payoff by multiplying the expected payoff of
each branch (shown in boldface to the right of the branch) by the probability of that branch
and then summing these products. Record this expected payoff for each deci- sion fork in
boldface next to the fork, and designate this quantity as also being the expected payoff for
the branch leading to this fork.
3. For each decision fork, compare the expected payoffs of its branches and choose the
alternative whose branch has the largest expected payoff. In each case, record the choice
on the decision tree by inserting a double dash as a barrier through each rejected branch.
To begin the procedure, consider the rightmost column of forks, namely, chance forks
𝑓𝑓, 𝑔𝑔, and ℎ. Applying step 2, their expected payoffs (EP) are calculated as,
1 6
𝐸𝐸𝐸𝐸 = 670 + −130 = −15.7, for fork 𝑓𝑓.
7 7
1 1
𝐸𝐸𝐸𝐸 = 670 + −130 = 270, for fork 𝑔𝑔.
2 2
1 3
𝐸𝐸𝐸𝐸 = 700 + −100 = 100, for fork ℎ.
4 4

4747
Click to edit
Decision trees
Master title style
Next, we move one column to the left, which consists of decision forks 𝑐𝑐, 𝑑𝑑, and 𝑒𝑒. The
expected payoff for a branch that leads to a chance fork now is recorded in boldface over that
chance fork. Therefore, step 3 can be applied as follows.
Fork 𝑐𝑐: Drill the alternative has 𝐸𝐸𝐸𝐸 = −15.7.
Sell the alternative has 𝐸𝐸𝐸𝐸 = 60.
60 > −15.7, so choose the sell alternative.
Fork 𝑑𝑑: Drill the alternative has 𝐸𝐸𝐸𝐸 = 270.
Sell the alternative has 𝐸𝐸𝐸𝐸 = 60.
270 > 60, so choose the drill alternative.
Fork 𝑒𝑒: Drill the alternative has 𝐸𝐸𝐸𝐸 = 100.
Sell the alternative has 𝐸𝐸𝐸𝐸 = 90.
100 > 90, so choose the drill alternative.
The expected payoff for each chosen alternative now would be recorded in boldface over its
decision node, as already shown in the figure (next slide). 4848
Click to edit
Decision trees
Master title style

4949
Click to edit
Decision trees
Master title style
Next, moving one more column to the left brings us to fork 𝑏𝑏. Since this is a chance fork, step 2
of the procedure needs to be applied. The expected payoff for each of its branches is recorded
over the following decision fork. Therefore, the expected payoff is,
𝐸𝐸𝐸𝐸 = 0.7 60 + 0.3 270 = 123, for fork 𝑏𝑏.
Finally, we move left to fork 𝑎𝑎, a decision fork. Applying step 3 yields,
Fork 𝑎𝑎: Do seismic survey has 𝐸𝐸𝐸𝐸 = 123.
No seismic survey has 𝐸𝐸𝐸𝐸 = 100.
123 > 100, so do seismic survey.
This procedure has moved from right to left for analysis purposes. However, having completed
the decision tree in this way, the decision maker now can read the tree from left to right to see
the actual progression of events. The double dashes have closed off the undesirable paths.
Following the open paths from left to right in the figure yields the following optimal policy,
according to Bayes’ decision rule.
Optimal policy: Do the seismic survey. If the result is unfavourable, sell the land. If the result is
favourable, drill for oil. The expected payoff (including the cot of seismic survey is 123 5050
($123000).
Click totheory
Utility edit Master title style
Thus far, when applying Bayes’ decision rule, we have assumed that the expected payoff in
monetary terms is the appropriate measure of the consequences of taking an action. However,
in many situations this assumption is inappropriate.
For example, suppose that an individual is offered the choice of (1) accepting a 50:50 chance
of winning $100,000 or nothing or (2) receiving $40,000 with certainty. Many people would
prefer the $40,000 even though the expected payoff on the 50:50 chance of winning $100,000
is $50,000.
A company may be unwilling to invest a large sum of money in a new product even when the
expected profit is substantial if there is a risk of losing its investment and thereby becoming
bankrupt.
People buy insurance even though it is a poor investment from the viewpoint of the expected
payoff.
Do these examples invalidate Bayes’ decision rule? Fortunately, the answer is no, because
there is a way of transforming monetary values to an appropriate scale that reflects the
decision maker’s preferences. This scale is called the utility function for money.

5151
Click totheory
Utility edit Master title style
Utility Functions for Money: The following figure shows a typical utility function 𝑢𝑢(𝑀𝑀) for
money 𝑀𝑀. It indicates that an individual having this utility function would value obtaining
$30,000 twice as much as $10,000 and would value obtaining $100,000 twice as much as
$30,000.

5252
Click totheory
Utility edit Master title style
This reflects the fact that the person’s highest-priority needs would be met by the first $10,000.
Having this decreasing slope of the function as the amount of money increases is referred to
as having a decreasing marginal utility for money. Such an individual is referred to as being
risk-averse.
However, not all individuals have a decreasing marginal utility for money. Some people are
risk seekers instead of risk-averse, and they go through life looking for the “big score.” The
slope of their utility function increases as the amount of money increases, so they have an
increasing marginal utility for money.
The intermediate case is that of a risk-neutral individual, who prizes money at its face value.
Such an individual’s utility for money is simply proportional to the amount of money involved.
Although some people appear to be risk-neutral when only small amounts of money are
involved, it is unusual to be truly risk-neutral with very large amounts.
It also is possible to exhibit a mixture of these kinds of behaviour. For example, an individual
might be essentially risk-neutral with small amounts of money, then become a risk seeker with
moderate amounts, and then turn risk-averse with large amounts.

5353
 Utility Curve
Click to edit Master title style

Risk Neutral
Utility
Risk Averse

Risk Seeking

Monetary Value

54
Click totheory
Utility edit Master title style
An individual’s attitude toward risk also may be different when dealing with one’s personal
finances than when making decisions on behalf of an organization. For example, managers of
a business firm need to consider the company’s circumstances and the collective philosophy of
top management in determining the appropriate attitude toward risk when making managerial
decisions.
When a utility function for money is incorporated into a decision analysis approach to a
problem, this utility function must be constructed to fit the preferences and values of the
decision maker involved. (The decision maker can be either a single individual or a group of
people.)
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
alternative courses of action if the two alternatives have the same expected utility.
To illustrate, suppose that the decision maker has the utility function shown in the figure.
Further suppose that the decision maker is offered the following opportunity.
Offer: An opportunity to obtain either $100,000 (utility= 4) with probability 𝑝𝑝 or nothing (utility =
0) with probability (1 − 𝑝𝑝).
5555
Click totheory
Utility edit Master title style
Thus,
𝐸𝐸 utility = 4𝑝𝑝, for this offer.
Therefore, for each of the following three pairs of alternatives, the decision maker is indifferent
between the first and second alternatives (refer to the figure):
The offer with 𝑝𝑝 = 0.25 (𝐸𝐸[utility]= 1), or definitely obtaining $10,000 (utility= 1).
The offer with 𝑝𝑝 = 0.5 (𝐸𝐸[utility]= 2), or definitely obtaining $30,000 (utility= 2).
The offer with 𝑝𝑝 = 0.75 (𝐸𝐸[utility]= 3), or definitely obtaining $10,000 (utility= 3).
The scale of the utility function (e.g., utility = 1 for $10,000) is irrelevant. It is only the relative
values of the utilities that matter. All the utilities can be multiplied by any positive constant
without affecting which alternative course of action will have the largest expected utility.
When the decision maker’s utility function for money is used to measure the relative worth of
the various possible monetary outcomes, Bayes’ decision rule replaces monetary payoffs by
the corresponding utilities. Therefore, the optimal action (or series of actions) is the one which
maximizes the expected utility.

5656
Click totheory
Utility edit Master title style
Applying Utility Theory to the Goferbroke Co. Problem
We mentioned that the Goferbroke Co. was operating without much capital, so a loss of
$100,000 would be quite serious. The (primary) owner of the company already has gone
heavily into debt to keep going. The worst-case scenario would be to come up with $30,000 for
a seismic survey and then still lose $100,000 by drilling when there is no oil. This scenario
would not bankrupt the company at this point, but definitely would leave it in a precarious
financial position. On the other hand, striking oil is an exciting prospect, since earning
$700,000 finally would put the company on a fairly solid financial footing.
To apply the owner’s (decision maker’s) utility function for money to the problem, it is
necessary to identify the utilities for all the possible monetary payoffs. In units of thousands of
dollars, these possible payoffs and the corresponding utilities are given in the following table.
Monetary payoff Utility Monetary payoff Utility
-130 -150 90 90
-100 -105 670 580
60 60 700 600
5757
Click totheory
Utility edit Master title style
As a starting point in constructing the utility function, it is natural to let the utility of zero money
be zero, so 𝑢𝑢 0 = 0. An appropriate next step is to consider the worst scenario and best
scenario and then to address the following question: Suppose you have only the following two
alternatives. Alternative 1 is to do nothing (payoff and utility = 0). Alternative 2 is to have a
probability 𝑝𝑝 of a payoff of 700 and a probability 1 − 𝑝𝑝 of a payoff of −130(loss of 130). What
value of p makes you indifferent between two alternatives?
1
The decision maker’s choice: 𝑝𝑝 = 5 .

If we continue to let 𝑢𝑢(𝑀𝑀) denote the utility of a monetary payoff of 𝑀𝑀, this choice of 𝑝𝑝 implies
that,
4 1
𝑢𝑢 −130 + 𝑢𝑢 700 = 0 (utility of alternative I)
5 5

The value of either 𝑢𝑢(−130) or 𝑢𝑢(700) can be set arbitrarily (provided only that the first is
negative and the second positive) to establish the scale of the utility function. By choosing
𝑢𝑢 −130 = −150 (a convenient choice since it will make 𝑢𝑢(𝑀𝑀) approximately equal to 𝑀𝑀 when
𝑀𝑀 is in the vicinity of 0), this equation then yields 𝑢𝑢 700 = 600.
5858
Click totheory
Utility edit Master title style
To identify 𝑢𝑢(−100), a choice of 𝑝𝑝 is made that makes the decision maker indifferent between a
payoff of −130 with probability 𝑝𝑝 or definitely incurring a payoff of −100. The choice is 𝑝𝑝 = 0.7,
so,
𝑢𝑢 −100 = 𝑝𝑝 𝑢𝑢 −130 = 0.7 −150 = −105.
To obtain 𝑢𝑢(90), a choice of 𝑝𝑝 is made that makes the decision maker indifferent between a
payoff of 700 with probability 𝑝𝑝 or definitely incurring a payoff of 90. The choice is 𝑝𝑝 = 0.15, so,
𝑢𝑢 90 = 𝑝𝑝 𝑢𝑢 700 = 0.15 600 = 90.
At this point, a smooth curve was drawn through 𝑢𝑢(−130), 𝑢𝑢(−100), 𝑢𝑢(90), and 𝑢𝑢(700) to obtain
the decision maker’s utility function for money shown in the figure (next slide). The values on
this curve at 𝑀𝑀 = 60 and 𝑀𝑀 = 670 provide the corresponding utilities, 𝑢𝑢 60 = 60 and 𝑢𝑢 670 =
580, which completes the list of utilities.
For contrast, the dashed line drawn at 45° in the figure shows the monetary value 𝑀𝑀 of the
amount of money 𝑀𝑀. This dashed line has provided the values of the payoffs used exclusively
in the preceding sections. Note how 𝑢𝑢(𝑀𝑀) essentially equals 𝑀𝑀 for small values (positive or
negative) of 𝑀𝑀, and then how 𝑢𝑢(𝑀𝑀) gradually falls off M for larger values of M. This is typical for
a moderately risk-averse individual.
5959
Click totheory
Utility edit Master title style

6060
Click totheory
Utility edit Master title style

Using a Decision Tree to Analyze the Goferbroke Co. Problem with Utilities
Now that the utility function for money of the owner of the Goferbroke Co. has been obtained,
this information can be used with a decision tree as summarized next.
The approach used in the preceding sections of maximizing the expected monetary payoff
amounts to assuming that the decision maker is risk-neutral, so that 𝑢𝑢 𝑀𝑀 = 𝑀𝑀. By using utility
theory, the optimal solution now reflects the decision maker’s attitude about risk. Because the
owner of the Goferbroke Co. adopted only a moderately risk-averse stance, the optimal policy
did not change from before. For a somewhat more risk-averse owner, the optimal solution
would switch to the more conservative approach of immediately selling the land (no seismic
survey).

6161
Click totheory
Utility edit Master title style

6262
Click to edit Master title style

63
 G: Growth
Click to edit Master title style
R:
Recession
G
D: R
S G
Depression
S : Stocks
B
B : Bonds R
G G
S R
R D
B G
S R
D
G
S R
G
B B
G R
G
R
R S D
B G
R
D
64
 G R G 144

Click toG:edit
Growth
Master title style 120
0.7
0.3
0.2
0.7 S R 108
G 126
R: Recession
0.0 0.1 B
D: Depression 108 R 132
G G
S : Stocks R 81
S
B : Bonds R D 45 94.5
B G
S 90 R 99

100 D 108
G 126

105 R 94.5
S
G 110.25
B B
G 132
R 115.5
G
Eco. Env. Rate of Return R 99
R S D 55
Stocks Bonds
B 115.5
Growth 20% 5% 110 G
R 121
Recession -10% 10%
D 132
Depression -50% 20%
65
 G R 133 0.7
G 144

Click to edit Master title style

G: Growth 0.7
0.3
0.2
0.7
133 S 0.3
R 108 128 0.7
G 126
R: Recession
0.0 0.1 B
D: Depression 108 0.3
R 132
123 0.7
G 83 G
0.2
S : Stocks R
0.7 81
S
B : Bonds 0.3
R D
0.1 45 94.5
B 99 G
0.2
S 99 R
0.7 99
D
0.1 108
117 0.7
G 126
123
117 S 0.3
R 94.5
112 0.7
G 110.25
B B
118 0.7
G 132
R
0.3 115.5
101 0.2
G
0.7
R 99
Eco. Env. Rate of Return 121 0.1
0.3
R S D 55
Stocks Bonds
B 121 0.2 115.5
Growth 20% 5% G
0.7
R 121
Recession -10% 10% 0.1
D 132
Depression -50% 20%
66
Click to edit Master title style

Applications and
conclusions

6767
Clickpractical
The to edit Master
application
title style
of decision analysis
• In one sense, this chapter’s prototype example (the Goferbroke Co. problem) is a very
typical application of decision analysis. Like other applications, management needed to
make some decisions (Do a seismic survey? Drill for oil or sell the land?) in the face of great
uncertainty. The decisions were difficult because their payoffs were so unpredictable.
• The outcome depended on factors that were outside management’s control (does the land
contain oil or is it dry?). Therefore, management needed a framework and methodology for
rational decision making in this uncertain environment.
• In other ways, the Goferbroke problem is not such a typical application. It was oversimplified
to include only two possible states of nature (Oil and Dry), whereas there actually would be a
considerable number of distinct possibilities. Management also was considering only two
alternatives for each of two decisions. Real applications commonly involve more decisions,
more alternatives to be considered for each one, and many possible states of nature.
• When dealing with larger problems, the decision tree can explode in size, with perhaps many
thousand terminal branches. In this case, it clearly would not be feasible to construct the tree
by hand, including computing posterior probabilities, and calculating the expected payoffs (or
utilities) for the various forks, and then identifying the optimal decisions. Fortunately, some
excellent software packages (mainly for personal computers) are available specifically for 6868
doing this work.
Clickpractical
The to edit Master
application
title style
of decision analysis
• Sensitivity analysis also can become unwieldy on large problems. Although it normally is
supported by the computer software, the amount of data generated can easily overwhelm an
analyst or decision maker. Therefore, some graphical techniques, such as tornado diagrams,
have been developed to organize the data in a readily understandable way.
• Many strategic business decisions are made collectively by several members of
management. One technique for group decision making is called decision conferencing. This
is a process where the group comes together for discussions in a decision conference with
the help of an analyst and a group facilitator. With the assistance of a computerized group
decision support system, the analyst builds and solves models on the spot, and then
performs sensitivity analysis to respond to what-if questions from the group.
• Applications of decision analysis commonly involve a partnership between the managerial
decision maker (whether an individual or a group) and an analyst (whether an individual or a
team) with training in OR.
• The following table (next slide) briefly summarizes the nature of some of the applications of
decision analysis.

6969
Clickpractical
The to edit Master
application
title style
of decision analysis
Organisation Nature of application

Amoco Oil Co. Used utilities to evaluate strategies for marketing its products through full-
facility service stations.
Ohio Edison Co. Evaluated and selected particulate emission control equipment for a coal-
fired power plant.
New England Electric System Determined an appropriate bid for the salvage rights to a grounded ship.

National weather service Developed a plan for responding to flood forecasts and warnings.

National Forest Administrations Planned prescribed fires to improve forest and rangeland ecosystems.

Tomco oil Corp. Chose between two site locations for drilling an oil well, with 74 states of
nature.
Personal Decision Used decision criteria without probabilities to choose between adjustable-
rate and fixed-rate mortgages.
U.S. Postal Service Chose between six alternatives for a postal automation program, saving
$200 million.
Santa Clara University Evaluated whether to implement a drug-testing program for their
intercollegiate athletes.
Independent Living Center (Australia) A decision conference developed a strategic plan for reorganizing the
centre.
DuPont Corp. Many applications to strategic planning; one added $175 million in value.
7070
Click to edit Master title style
Conclusions
• Decision analysis has become an important technique for decision making in the face of
uncertainty. It is characterized by enumerating all the available courses of action, identifying
the payoffs for all possible outcomes, and quantifying the subjective probabilities for all the
possible random events. When these data are available, decision analysis becomes a
powerful tool for determining an optimal course of action.
• One option that can be readily incorporated into the analysis is to perform experimentation to
obtain better estimates of the probabilities of the possible states of nature. Decision trees are
a useful visual tool for analysing this option or any series of decisions.
• Utility theory provides a way of incorporating the decision maker’s attitude toward risk into
the analysis.

7171