DECISION THEORY

What makes the difference?

90% 10%

Good Decision Bad Decision

A “good” decision – one that uses analytic decision making – is based on logic
and considers all available data and possible alternatives.
• A good decision
follows these steps:

1. Clearly define the problem and

the factors that influence it.

2. Develop specific and

measurable objectives.

3. Develop a model – that is, a

relationship between objectives
and variables (which are
measurable quantities).
• A good decision
follows these steps:

4. Evaluate each alternative

solution based on its merits and
drawbacks.

5. Select the best alternative.

6. Implement and evaluate the

decision and then set a
timetable for completion.
DECISION
TREE
FUNDAMENTALS OF
DECISION MAKING

Regardless of the complexity of a decision or the sophistication of the

technique used to analyze it, all decision makers are faced with
alternatives and “states of nature”.

Terms:

Alternative – A course of action or strategy that may be chosen by a decision

maker.

State of nature – An occurrence or a situation over which the decision maker

has little or no control.
FUNDAMENTALS OF
DECISION MAKING

Regardless of the complexity of a decision or the sophistication of the

technique used to analyze it, all decision makers are faced with
alternatives and “states of nature”.

Symbols used in a decision tree:

– Decision node from which one of several alternatives may be selected.

– A state-of-nature node out of which one state of nature will occur.

FUNDAMENTALS OF
DECISION MAKING

To present a manager’s decision alternatives, we can develop decision trees

using the above symbols.

When constructing a decision tree, we must be sure that all alternatives

and states of nature are in their correct and logical places and that we
include all possible alternatives and states of nature, USUALLY
INCLUDING THE “DO NOTHING” OPTION.
DECISION TREE

 A decision tree is a chronological representation of the decision problem.

 Each decision tree has two types of nodes: round nodes correspond to
the states of nature while square nodes correspond to the decision
alternatives.

 The branches leaving each round node represent the different states
of nature while the branches leaving each square node represent the
different decision alternatives.

 At the end of each limb of a tree are the payoffs attained from the
series of branches making up that limb.
DECISION TREE
Example:

Getz Products Company is investigating the possibility of producing and

marketing backyard storage sheds. Undertaking this project would
require the construction of either a large or a small manufacturing plant.
The market for the product produced—storage sheds—could be either
favorable or unfavorable. Getz, of course, has the option of not
developing the new product line at all.

Getz decides to build a decision tree.

DECISION
TABLE
DECISION TABLE

 Decision table —A tabular means of analyzing decision alternatives and

states of nature.

 A decision table is sometimes called a payoff table. For any alternative

and a particular state of nature, there is a consequence, or an outcome ,
which is usually expressed as a monetary value; this is called the
conditional value.
DECISION TABLE
Example:

Getz Products now wishes to organize the following information into a

table. With a favorable market, a large facility will give Getz Products a
net profit of $200,000. If the market is unfavorable, a $180,000 net
loss will occur. A small plant will result in a net profit of $100,000 in a
favorable market, but a net loss of $20,000 will be encountered if the
market is unfavorable.

These numbers become conditional values in the decision table. We list

alternatives in the left column and states of nature across the top of
the table.
 TYPES OF
DECISION-
MAKING
ENVIRONMENTS
 TYPES OF DECISION-MAKING ENVIRONMENTS

The types of decisions people make depend on how much

knowledge or information they have
about the situation. There are three decision-making environments:

Decision Making under Decision Making under Risk. Decision Making under
Uncertainty. Decision maker has some Certainty.
Decision maker has no knowledge regarding Decision maker know for
information at all probability of occurrence sure (that is, with
about various of each outcome or state certainty) outcome or
outcomes or states of of nature. consequence of every
nature. decision alternative.
 DECISION
MAKING UNDER
UNCERTAINTY
DECISION MAKING UNDER
UNCERTAINTY

 If the decision maker does not know with certainty which state of nature
will occur, then he/she is said to be making decision under uncertainty.

 When there is complete uncertainty as to which state of nature in a

decision environment may occur (i.e., when we cannot even assess
probabilities for each possible outcome), we rely on three decision
methods:

Maximax: This method finds an alternative that maximizes the maximum

outcome for every alternative.

First, we find the maximum outcome within every alternative, and then
we pick the alternative with the maximum number. Because this
decision criterion locates the alternative with the highest possible gain,
it has been called an “optimistic” decision criterion.
DECISION MAKING UNDER
UNCERTAINTY

 If the decision maker does not know with certainty which state of nature
will occur, then he/she is said to be making decision under uncertainty.

 When there is complete uncertainty as to which state of nature in a

decision environment may occur (i.e., when we cannot even assess
probabilities for each possible outcome), we rely on three decision
methods:

Maximin: This method finds the alternative that maximizes the minimum
outcome for every alternative.

First, we find the minimum outcome within every alternative, and then we
pick the alternative with the maximum number. Because this decision
criterion locates the alternative that has the least possible loss, it has
been called a “pessimistic” decision criterion.
DECISION MAKING UNDER
UNCERTAINTY

 If the decision maker does not know with certainty which state of nature
will occur, then he/she is said to be making decision under uncertainty.

 When there is complete uncertainty as to which state of nature in a

decision environment may occur (i.e., when we cannot even assess
probabilities for each possible outcome), we rely on three decision
methods:

Equally likely: This method finds the alternative with the highest average
outcome.

First, we calculate the average outcome for every alternative, which is the
sum of all outcomes divided by the number of outcomes. We then
pick the alternative with the maximum number. The equally likely
approach assumes that each state of nature is equally likely to occur.
DECISION MAKING UNDER
UNCERTAINTY
Example:

Getz Products now wishes to organize the following information into a

table. With a favorable market, a large facility will give Getz Products a
net profit of $200,000. If the market is unfavorable, a $180,000 net
loss will occur. A small plant will result in a net profit of $100,000 in a
favorable market, but a net loss of $20,000 will be encountered if the
market is unfavorable.

These numbers become conditional values in the decision table. We list

alternatives in the left column and states of nature across the top of
the table.
DECISION MAKING UNDER
UNCERTAINTY

Example:

Given Getz’s decision table, we will now determine the maximax,

maximin, and equally likely decision criteria.
DECISION MAKING UNDER
UNCERTAINTY
Example:

1. The maximax choice is to construct a large plant. This is the maximum

of the maximum number within each row, or alternative.

2. The maximin choice is to do nothing. This is the maximum of the

minimum within each row, or alternative.

3. The equally likely choice is to construct a small plant. This is the

maximum of the average outcome of each alternative. This approach
assumes that all outcomes for any alternative are equally likely.

There are optimistic decision makers (“maximax”) and pessimistic ones

(“maximin”).

Maximax and maximin present the best case – worst case planning
scenarios.
 DECISION
MAKING UNDER
RISK
DECISION MAKING UNDER
RISK

 Decision making under risk, a more common occurrence, relies on

probabilities.

 Several possible states of nature may occur, each with an assumed

probability.

 The states of nature must be mutually exclusive and collectively

exhaustive and their probabilities must sum to 1.

If probabilistic information regarding the states of nature is

available, one may use the expected Monetary value
(EMV) approach (also known as Expected Value or EV).
DECISION MAKING UNDER
RISK

 Given a decision table with conditional values and probability

assessments for all states of nature, we can determine the expected
monetary value (EMV) for each alternative – the expected value or
mean return for each alternative if we could repeat this decision (or
similar types of decisions) a large number of times.

 Here the expected return for each decision is calculated by summing

the products of the payoff under each state of nature and the probability
of the respective state of nature occurring.

 The decision yielding the best expected return is chosen.

The expected value of a decision alternative is the sum of

weighted payoffs for the decision alternative.
DECISION MAKING UNDER
RISK

 The EMV for an alternative is the sum of all possible payoffs from the
alternative, each weighted by the probability of that payoff occurring:

N
EV( di )   P( s j )Vij
j 1 OR
where: N = the number of states of nature
P(sj) = the probability of state of
nature sj
Vij = the payoff corresponding to
decision alternative di and state of
nature sj
DECISION MAKING UNDER
RISK
Example:

Getz Products now wishes to organize the following information into a

table. With a favorable market, a large facility will give Getz Products a
net profit of $200,000. If the market is unfavorable, a $180,000 net
loss will occur. A small plant will result in a net profit of $100,000 in a
favorable market, but a net loss of $20,000 will be encountered if the
market is unfavorable.

These numbers become conditional values in the decision table. We list

alternatives in the left column and states of nature across the top of
the table.
DECISION MAKING UNDER
RISK

Example:

Getz Products’ operations manager believes that the probability of a

favorable market is 0.6, and that of an unfavorable market is 0.4. We can
now determine the EMV for each alternative.
DECISION MAKING UNDER
RISK

Example:

The maximum EMV is seen in alternative A2 . Thus, according to the EMV

decision criterion, Getz would build the small facility.
 DECISION
MAKING UNDER
CERTAINTY
DECISION MAKING UNDER
CERTAINTY

 If a manager were able to determine which state of nature would occur,

then he or she would know which decision to make. Once a manager
knows which decision to make, the payoff increases because the
payoff is now a certainty, not a probability. Because the payoff will
increase with knowledge of which state of nature will occur, this
knowledge has value.

 Therefore, we now look at how to determine the value of this

information.

 We call this difference between the payoff under perfect information

and the payoff under risk the expected value of perfect information
(EVPI).
DECISION MAKING UNDER
CERTAINTY

 Frequently information is available that can improve the probability

estimates for the states of nature.

 The expected value of perfect information (EVPI) is the increase in the

expected profit that would result if one knew with certainty which state
of nature would occur.

 The EVPI provides an upper bound on the expected value of any sample
or survey information. It places an upper bound on what you would be
willing to spend on information, such as that being sold by a marketing
consultant.
DECISION MAKING UNDER
CERTAINTY

 To find the EVPI, we must first compute the expected value with
perfect information (EVwPI), which is the expected (average) return if
we have perfect information before a decision has to be made.

 To calculate this value, we choose the best alternative for each state of
nature and multiply its payoff times the probability of occurrence of that
state of nature:
DECISION MAKING UNDER
CERTAINTY
Example:

Now suppose that the Getz operations manager has been approached by a marketing
research firm that proposes to help him make the decision about whether to build the
plant to produce storage sheds. The marketing researchers claim that their technical
analysis will tell Getz with certainty whether the market is favorable for the proposed
product. In other words, it will change Getz’s environment from one of decision
making under risk to one of decision making under certainty . This information could
prevent Getz from making a very expensive mistake. The marketing research firm
would charge Getz $65,000 for the information.

What would you recommend? Should the operations manager hire the firm to make
the study? Even if the information from the study is perfectly accurate, is it worth
$65,000? What might it be worth?
DECISION MAKING UNDER
CERTAINTY
Example:

The Getz operations manager would like to calculate the maximum that he would pay
for information— that is, the expected value of perfect information, or EVPI.

We follow a two-stage process. First, the expected value with perfect information
(EVwPI) is computed. Then, using this information, the EVPI is calculated.

The best outcome for the state of nature “favorable market” is “build a large facility”
with a payoff of $200,000. The best outcome for the state of nature “unfavorable
market” is “do nothing” with a payoff of $0.

Expected value with perfect information = ($200,000)(0.6) + ($0)(0.4) = $120,000.

Thus, if we had perfect information, we would expect (on the average) $120,000 if
the decision could be repeated many times.
DECISION MAKING UNDER
CERTAINTY
Example:

The maximum EMV is $52,000 for A 2 , which is the expected outcome without
perfect information.
DECISION MAKING UNDER
CERTAINTY
Example:

Thus:

The most Getz should be willing to pay for perfect information is $68,000. This
conclusion, of course, is again based on the assumption that the probability of the first
state of nature is 0.6 and the second is 0.4.
DECISION TREE
DECISION TREE

 Decisions that lend themselves to display in a decision table also lend

themselves to display in a decision tree.

 Although the use of a decision table is convenient in problems having

one set of decisions and one set of states of nature, many problems
include sequential decisions and states of nature.

 When there are two or more sequential decisions, and later decisions
are based on the outcome of prior ones, the decision tree approach
becomes appropriate.

A decision tree is a graphic display of the decision process that

indicates decision alternatives, states of nature and their
respective probabilities, and payoffs for each combination
of decision alternative and state of nature.
DECISION TREE

 Expected monetary value (EMV) is the most commonly used criterion for
decision tree analysis.

 One of the first steps in such analysis is to graph the decision tree and to
specify the monetary consequences of all outcomes for a particular
problem.

A decision tree is a graphic display of the decision process that

indicates decision alternatives, states of nature and their
respective probabilities, and payoffs for each combination
of decision alternative and state of nature.
DECISION TREE

 Analyzing problems with decision trees involves five (5) steps:

1. Define the problem.

2. Structure or draw the decision tree.
3. Assign probabilities to the states of nature.
4. Estimate payoffs for each possible combination of decision
alternatives and states of nature.
5. Solve the problem by computing the expected monetary values
(EMV) for each state-of-nature node. This is done by working
backward —that is, by starting at the right of the tree and working
back to decision nodes on the left.

A decision tree is a graphic display of the decision process that

indicates decision alternatives, states of nature and their
respective probabilities, and payoffs for each combination
of decision alternative and state of nature.
DECISION TREE

Example:

Getz wants to develop a completed and solved decision tree.

DECISION TREE

Example:

The payoffs are placed at the right-hand side of each of the tree’s branches The
probabilities are placed in parentheses next to each state of nature. The expected
monetary values for each state-of-nature node are then calculated and placed by
their respective nodes.

The EMV of the first node is $48,000. This represents the branch from the decision
node to “construct a large plant.”

The EMV for node 2, to “construct a small plant,” is $52,000.

The option of “doing nothing” has, of course, a payoff of $0.

DECISION TREE
Example:

The branch leaving the decision node leading to the state-of-nature node with the
highest EMV will be chosen. In Getz’s case, a small plant should be built.
COMPLEX DECISION TREE

 Note that all possible outcomes and alternatives are included in their
logical sequence. This procedure is one of the strengths of using
decision trees. The manager is forced to examine all possible outcomes,
including unfavorable ones. He or she is also forced to make decisions in
a logical, sequential manner.

A decision tree is a graphic display of the decision process that

indicates decision alternatives, states of nature and their
respective probabilities, and payoffs for each combination
of decision alternative and state of nature.
COMPLEX DECISION TREE

Example:

Getz Products wishes to develop the new tree for this sequential decision.
COMPLEX DECISION TREE

Example:

Assuming Getz Products has two decisions to make, with the second decision
dependent on the outcome of the first.

Before deciding about building a new plant, Getz has the option of conducting its own
marketing research survey, at a cost of $10,000. The information from this survey
could help it decide whether to build a large plant, to build a small plant, or not to
build at all. Getz recognizes that although such a survey will not provide it with perfect
information, it may be extremely helpful.
COMPLEX DECISION TREE

Example:
COMPLEX DECISION TREE

Example:

Examining the tree, we see that Getz’s first decision point is whether to conduct the
$10,000 market survey. If it chooses not to do the study (the lower part of the tree), it
can either build a large plant, a small plant, or no plant. This is Getz’s second decision
point. If the decision is to build, the market will be either favorable (0.6 probability) or
unfavorable (0.4 probability). The payoffs for each of the possible consequences are
listed along the right-hand side.

The upper part of the tree reflects the decision to conduct the market survey. State-
of-nature node number 1 has 2 branches coming out of it. Let us say there is a 45%
chance that the survey results will indicate a favorable market for the storage sheds.
We also note that the probability is 0.55 that the survey results will be negative.
COMPLEX DECISION TREE

Example:

The rest of the probabilities shown in parentheses in the tree are all conditional
probabilities.

For example, 0.78 is the probability of a favorable market for the sheds given a
favorable result from the market survey. Of course, you would expect to find a high
probability of a favorable market given that the research indicated that the market
was good. Don’t forget, though: There is a chance that Getz’s $10,000 market survey
did not result in perfect or even reliable information. Any market research study is
subject to error. In this case, there remains a 22% chance that the market for sheds will
be unfavorable given positive survey results.
COMPLEX DECISION TREE

Example:

Likewise, we note that there is a 27% chance that the market for sheds will be
favorable given negative survey results. The probability is much higher, 0.73, that the
market will actually be unfavorable given a negative survey.

Finally, when we look to the payoff column in Figure A.3 , we see that $10,000—the
cost of the marketing study—has been subtracted from each of the top 10 tree
branches.

Thus, a large plant constructed in a favorable market would normally net a $200,000
profit. Yet because the market study was conducted, this figure is reduced by
$10,000. In the unfavorable case, the loss of $180,000 would increase to $190,000.
Similarly, conducting the survey and building no plant now results in a –$10,000
payoff.
COMPLEX DECISION TREE
Example:

With all probabilities and payoffs specified, we can start calculating the expected
monetary value of each branch. We begin at the end or right-hand side of the
decision tree and work back toward the origin. When we finish, the best decision will
be known.
COMPLEX DECISION TREE
Example:

With all probabilities and payoffs specified, we can start calculating the expected
monetary value of each branch. We begin at the end or right-hand side of the
decision tree and work back toward the origin. When we finish, the best decision will
be known.
COMPLEX DECISION TREE

Example:
QUIZ
QUIZ NO. 3
1. Stella Yan Hua is considering the possibility of opening a small dress shop on
Fairbanks Avenue, a few blocks from the university. She has located a good mall
that attracts students. Her options are to open a small shop, a medium-sized shop,
or no shop at all. The market for a dress shop can be good, average, or bad. The
probabilities for these three possibilities are .2 for a good market, .5 for an average
market, and .3 for a bad market. The net profit or loss for the medium-sized or
small shops for the various market conditions are given in the adjacent table.
Building no shop at all yields no loss and no gain. What do you recommend?
QUIZ NO. 3
2. T.S. Amer’s Ski Shop in Nevada has a 100-day season. T.S. has established the
probability of various store traffic, based on historical records of skiing conditions,
as indicated in the table to the right. T.S. has four merchandising plans, each
focusing on a popular name brand. Each plan yields a daily net profit as noted in
the table. He also has a meteorologist friend who, for a small fee, will accurately
tell tomorrow’s weather so T.S. can implement one of his four merchandising
plans.

a) What is the expected monetary value (EMV) under risk?

b) What is the expected value with perfect information (EVwPI)?
c) What is the expected value of perfect information (EVPI)?
QUIZ NO. 3

3. Louisiana is busy designing new lottery scratchoff games. In the latest game,
Bayou Boondoggle, the player is instructed to scratch off one spot: A, B, or C. A
can reveal “Loser,” “Win $1,” or “Win $50.” B can reveal “Loser” or “Take a Second
Chance.” C can reveal “Loser” or “Win $500.” On the second chance, the player is
instructed to scratch off D or E. D can reveal “Loser” or “Win $1.” E can reveal
“Loser” or “Win $10.” The probabilities at A are .9, .09, and .01. The probabilities at
B are .8 and .2. The probabilities at C are .999 and .001. The probabilities at D are .5
and .5. Finally, the probabilities at E are .95 and .05. Draw the decision tree that
represents this scenario. Use proper symbols and label all branches clearly.
Calculate the expected value of this game.
QUIZ NO. 3

4. Joseph Biggs owns his own ice cream truck and lives 30 miles from a Florida
beach resort. The sale of his products is highly dependent on his location and on
the weather. At the resort, his profit will be $120 per day in fair weather, $10 per
day in bad weather. At home, his profit will be $70 in fair weather and $55 in bad
weather. Assume that on any particular day, the weather service suggests a 40%
chance of foul weather.

a) Construct Joseph’s decision tree.

b) What decision is recommended by the expected value criterion?
QUIZ NO. 3

5. Dwayne Whitten, president of Whitten Industries, is considering whether to build

a manufacturing plant in north Texas. His decision is summarized in the following
table:

a) Construct a decision tree.

b) Determine the best strategy using expected monetary value (EMV).
c) What is the expected value of perfect information (EVPI)?
QUIZ NO. 3

6. Given the following conditional value table, determine the appropriate decision
under uncertainty using:

a) Maximax
b) Maximin
c) Equally likely
QUIZ NO. 3
7. Even though independent gasoline stations have been having a difficult time, Ian Langella
has been thinking about starting his own independent gasoline station. Ian’s problem is to
decide how large his station should be. The annual returns will depend on both the size of his
station and a number of marketing factors related to the oil industry and demand for gasoline.
After a careful analysis, Ian developed the following table:

For example, if Ian constructs a small station and the market is good, he will realize a profit of
$50,000.

a) Develop a decision table.

b) What is the maximax decision?
c) What is the maximin decision?
d) What is the equally likely decision?
e) Develop a decision tree. Assume each outcome is equally likely, then find the highest
EMV.
QUIZ NO. 3
8. Jerry Bildery’s factory is considering three approaches for meeting an expected
increase in demand. These three approaches are increasing capacity, using
overtime, and buying more equipment. Demand will increase either slightly (S),
moderately (M), or greatly (G). The profits for each approach under each possible
scenario are as follows:

Since the goal is to maximize, and Jerry is risk-neutral, he decides to use the
equally likely decision criterion to make the decision as to which approach to use.
According to this criterion, which approach should be used?
QUIZ NO. 3
9. Andrew Thomas, a sandwich vendor at Hard Rock Cafe’s annual Rockfest,
created a table of conditional values for the various alternatives (stocking
decision) and states of nature (size of crowd):

The probabilities associated with the states of nature are 0.3 for a big demand, 0.5
for an average demand, and 0.2 for a small demand.

a) Determine the alternative that provides Andrew the greatest expected

monetary value (EMV).
b) Compute the expected value of perfect information (EVPI).
QUIZ NO. 3
10. Howard Weiss, Inc., is considering building a sensitive new radiation scanning
device. His managers believe that there is a probability of 0.4 that the ATR Co. will
come out with a competitive product. If Weiss adds an assembly line for the
product and ATR Co. does not follow with a competitive product, Weiss’s
expected profit is $40,000; if Weiss adds an assembly line and ATR follows suit,
Weiss still expects $10,000 profit. If Weiss adds a new plant addition and ATR
does not produce a competitive product, Weiss expects a profit of $600,000; if
ATR does compete for this market, Weiss expects a loss of $100,000.

a) Determine the EMV of each decision.

b) Compute the expected value of perfect information.
 "I cannot teach ANYBODY
ANYTHING, I can only make
them think.”

— Socrates
Thanks!
Does anyone have any questions?

CREDITS: This presentation template was

created by Slidesgo, including icons by
Flaticon, infographics & images by Freepik