12/20/2016

Chapter 9

 Business analytics is about making better decisions

 Decision analysis
l can beb used
d to develop
d l an optimall
strategy:
 When a decision maker is faced with several decision
alternatives and an uncertain or risk‐filled pattern of future
events
 For example: The State of North Carolina used decision
analysis in evaluating whether to implement a medical
screening test to detect metabolic disorders in newborns
 A good decision analysis includes careful
consideration of risk

1
 12/20/2016

 Risk analysis helps to provide the probability

i f
information
ti about
b t the
th favorable
f bl as wellll as the
th
unfavorable outcomes that may occur
 Decision analysis considers problems that
involve reasonably few decision alternatives and
reasonably few possible future events
 Topics to be discussed under decision analysis:
 Payoff
P ff tables
t bl andd decision
d i i trees
t
 Sensitivity analysis
 Use of Bayes’ theorem
 Utility theory

 Payoff Tables
 Decision Trees

2
 12/20/2016

 The first step

p in the decision analysis
y p process
is problem formulation:
 Create verbal statement of the problem
 Identify the decision alternatives; the uncertain
future events, referred to as chance events; the
outcomes associated with each combination of
decision alternative and chance event outcome

Example: Construction project of Pittsburgh

Development Corporation
 PDC commissioned preliminary architectural
drawings for three different projects:
 One with 30 condominiums
 One with 60 condominiums
 One
O with
i h 90 condominiums
d i i
 The financial success of the project depends on:
 The size of the condominium complex
 The chance event concerning the demand for the
condominiums 6

3
 12/20/2016

 The statement of the PDC decision problem is to

select the size of the new luxury condominium project
that will lead to the largest profit given the
uncertainty concerning the demand for the
condominiums
 Given the statement of the problem, it is clear that
the decision is to select the best size for the
condominium complex
 PDC has
h the
th following
f ll i three
th d i i alternatives:
decision lt ti
 d1 = a small complex with 30 condominiums
 d2 = a medium complex with 60 condominiums
 d3 = a large complex with 90 condominiums

 In decision analysis, the possible outcomes for a

chance
h eventt are the
th states
t t off nature
t
 The states of nature are mutually exclusive (no
more than one can occur) and collectively
exhaustive (at least one must occur)
 Thus one and only one of the possible states of
nature will occur
 For PDC example: The chance event concerning
the demand for the condominiums has two
states of nature:
 s1 = strong demand for the condominiums
 s2 = weak demand for the condominiums
8

4
 12/20/2016

Payoff
y Tables
 Payoff is the outcome resulting from a specific
combination of a decision alternative and a state of
nature
 Payoff table is a table showing payoffs for all
combinations of decision alternatives and states of
nature

TABLE 9.1: PAYOFF TABLE FOR THE PDC

CONDOMINIUM PROJECT ($ MILLIONS)

10

5
 12/20/2016

 We will use the notation Vij to denote the

payoff associated with decision alternative i
and state of nature j
 From Table 9.1, V31 = 20 indicates that a
payoff of $20 million occurs if the decision is
to build a large complex (d3) and the strong
demand state of nature (s1) occurs

11

Decision Tree
 A decision tree provides a graphical
representation of the decision‐making
process
 Shows the natural or logical progression that
will occur over time

12

6
 12/20/2016

The topmost
payoff of 8
indicates that an
$8 million profit
is anticipated if
PDC constructs a
small
condominium
d i i
complex (d1) and
demand turns
out to be strong
(s1)

13

 The decision tree in Figure

g 9.1 shows:
9
 Four nodes, numbered 1–4
 Nodes are used to represent decisions and chance
events
 Squares are used to depict decision nodes, circles are
used to depict chance nodes
 Node 1 is a decision node, and nodes 2, 3, and 4 are
chance nodes

14

7
 12/20/2016

 The branches connect the nodes;; those

leaving the decision node correspond to the
decision alternatives
 The branches leaving each chance node
correspond to the states of nature
 The outcomes (payoffs) are shown at the end
of the states‐of‐nature branches

15

 Optimistic Approach
 Conservative Approach
 Minimax Regret Approach

8
 12/20/2016

 Decision analysis
y without p
probabilities is
appropriate in situations:
 In which a simple best‐case and worst‐case analysis is
sufficient
 Where the decision maker has little confidence in his
or her abilityy to assess the p
probabilities

17

Optimistic
p Approach
pp
 The optimistic approach evaluates each decision
alternative in terms of the best payoff that can occur
 The decision alternative that is recommended is the one
that provides the best possible payoff
 For minimization problems, this approach leads to
choosing the alternative with the smallest payoff

18

9
 12/20/2016

TABLE 9.2: MAXIMUM PAYOFF FOR EACH PDC

DECISION ALTERNATIVE

In the PDC problem, the optimistic approach would lead the decision
maker to choose the alternative corresponding to the largest profit
19

Conservative Approach
pp
 The conservative approach evaluates each decision
alternative in terms of the worst payoff that can occur
 The decision alternative recommended is the one that
provides the best of the worst possible payoffs
 For problems involving minimization (for example, when
the
h output measure is i cost),
) this
hi approach
h identifies
id ifi the h
alternative that will minimize the maximum payoff

20

10
 12/20/2016

TABLE 9.3: MINIMUM PAYOFF FOR EACH PDC DECISION

ALTERNATIVE

In the PDC problem, the conservative approach would lead the decision maker
to choose the alternative that maximizes the minimum possible profit that
could be obtained
21

Minimax Regret Approach

 Regret
R t isi the
th difference
diff b t
between th payoff
the ff
associated with a particular decision alternative
and the payoff associated with the decision that
would yield the most desirable payoff for a
given state of nature
 Regret is often referred to as opportunity loss
 Under the minimax regret approach,
approach one would
choose the decision alternative that minimizes
the maximum state of regret that could occur
over all possible states of nature

22

11
 12/20/2016

 Using equation below and the payoffs in

Table
bl 9.1, the
h regret associated d withh each
h
combination of decision alternative di and
state of nature sj is computed
 To compute the regret, subtract each entry in
a column from the largest
g entryy in the column

23

TABLE 9.4: OPPORTUNITY LOSS, OR REGRET, TABLE

FOR THE PDC CONDOMINIUM PROJECT ($ MILLIONS)

12
 12/20/2016

 The next step

p in applying
pp y g the minimax regret
g
approach is to list the maximum regret for
each decision alternative
 For the PDC problem, the alternative to
construct the medium condominium
complex with a corresponding maximum
complex,
regret of $6 million, is the recommended
minimax regret decision

25

TABLE 9.5: MAXIMUM REGRET FOR EACH PDC

DECISION ALTERNATIVE

13
 12/20/2016

 Expected Value Approach

 Risk Analysis
 Sensitivity Analysis

Expected
p Value Approach
pp
 The expected value (EV) of a decision alternative is the
sum of weighted payoffs for the decision alternative
 The weight for a payoff is the probability of the
associated state of nature and therefore the probability
that the payoff will occur

28

14
 12/20/2016

29

30

15
 12/20/2016

 Select the decision branch leading g to the

chance node with the best expected value
 The decision alternative associated with this
branch is the recommended decision
 In practice, obtaining precise estimates of the
probabilities for each state of nature is often
impossible, so historical data is preferred to
use for estimating the probabilities for the
different states of nature
31

Risk Analysis
 Risk
Ri k analysis
l i helps
h l th
the d i i
decision maker
k
recognize the difference between the expected
value of a decision alternative and the payoff
that may actually occur
 Decision alternative and a state of nature
combine to generate the payoff associated with
a decision
 Risk profile for a decision alternative shows the
possible payoffs along with their associated
probabilities

32

16
 12/20/2016

33

Sensitivity Analysis
 Sensitivity
Sensiti it analysis
anal sis determines howho changes in the
probabilities for the states of nature or changes in the
payoffs affect the recommended decision alternative.
 In many cases, the probabilities for the states of nature
and the payoffs are based on subjective assessments
 Sensitivity analysis helps the decision maker understand
which of these inputs are critical to the choice of the best
decision
dec s o aalternative
te at e
 If a small change in the value of one of the inputs causes a
change in the recommended decision alternative, the
solution to the decision analysis problem is sensitive to
that particular input

34

17
 12/20/2016

 Example: Suppose that, in the PDC problem, the

probability for a strong demand is revised to 0.2
0 2 and
the probability for a weak demand is revised to 0.8
 EV(d1 ) = 0.2 (8) + 0.8 (7) = 7.2
 EV(d2 ) = 0.2 (14) + 0.8 (5) = 6.8
 EV(d3 ) = 0.2 (20) + 0.8 (‐9) = ‐3.2
 With these probability assessments, the
recommended decision alternative is to construct a
small condominium complex (d1), with an expected
value of $7.2 million
 When the probability of strong demand is large, PDC
should build the large complex; when the probability
of strong demand is small, PDC should build the small
complex
35

 Expected Value of Sample Information

 Expected Value of Perfect Information

18
 12/20/2016

 Decision makers have the ability to collect

additional information about the states of
nature
 Additional information is obtained through
experiments designed to provide sample
information about the states of nature
 The preliminary or prior probability
assessmentst for
f the
th states
t t off nature
t th t are the
that th
best probability values available prior to
obtaining additional information
 Posterior probabilities are revised probabilities
after obtaining additional information
37

 Example:
p PDC management
g is considering
g a 6‐
month market research study designed to learn
more about potential market acceptance of the
PDC condominium project, anticipating two
results:
 Favorable report:
p A substantial number of the individuals
contacted express interest in purchasing a PDC
condominium
 Unfavorable report: Very few of the individuals contacted
express interest in purchasing a PDC condominium

38

19
 12/20/2016

39

40

20
 12/20/2016

41

42

21
 12/20/2016

43

44

22
 12/20/2016

45

46

23
 12/20/2016

 If the market research is favorable,, construct

the large condominium complex
 If the market research is unfavorable,
construct the medium condominium
complex

47

Expected Value of Sample Information

 From Figure 8.9 we can concludel d that
h theh
difference, 15.93 – 14.20 = 1.73, is the
expected value of sample information (EVSI)

48

24
 12/20/2016

Expected
p Value of Perfect Information
 A special case of gaining additional information related
to a decision problem is when the sample information
provides perfect information on the states of nature

49

TABLE 8.6: PAYOFF TABLE FOR THE PDC

CONDOMINIUM PROJECT ($ MILLIONS)

We can state PDC’s optimal decision strategy when the perfect

information becomes available as follows:
 If s1, select d3 and receive a payoff of $20 million
 If s2, select d1 and receive a payoff of $7 million
50

25
 12/20/2016

 The original probabilities for the states of nature:

P(s1) = 0.8 and P(s2) = 0.2
 The expected value of the decision strategy that uses
perfect information is 0.8(20) + 0.2(7) = 17.4 (i.e.,
expected value with perfect information (EVwPI))
 Earlier, we found the expected value approach is
decision alternative d3 $14.2
$14 2 million; this is referred to as
the expected value without perfect information (EVwoPI)

51

 Example
p for PDC: Expectedp value of the
perfect information (EVPI) is $17.4 – $14.2 =
$3.2 million
 In general, expected value for perfect
information (EVPI) is computed as:

26
 12/20/2016

 Bayes’ theorem can be used to compute branch

probabilities for decision trees
 The notation | in P(s1|F) and P(s2|F) is read as “given” and
indicates a conditional probability because we are
interested in the probability of a particular state of
nature “conditioned” on the fact that we receive a
p
favorable market report
 P(s1|F) and P(s2|F) are referred to as posterior
probabilities because they are conditional probabilities
based on the outcome of the sample information

54

27
 12/20/2016

55

56

28
 12/20/2016

 In p
performing g the p
probabilityy computations,
p
we need to know PDC’s assessment of the
probabilities of the two states of nature, P(s1)
and P(s2)
 We must know the conditional probability of
the market research outcomes given each
state off nature
 To carry out the probability calculations, we
need conditional probabilities for all sample
outcomes given all states of nature
57

 In the PDC problem we assume that the following

assessments are available for these conditional
probabilities:

 Bayes’ Theorem restated is:

58

29
 12/20/2016

TABLE 9.7: BRANCH PROBABILITIES FOR THE PDC

CONDOMINIUM PROJECT BASED ON A FAVORABLE
MARKET RESEARCH REPORT

59

TABLE 9.8: BRANCH PROBABILITIES FOR THE PDC

CONDOMINIUM PROJECT BASED ON AN UNFAVORABLE
MARKET RESEARCH REPORT

60

30
 12/20/2016

 Utility and Decision Analysis

 Utility Functions
 Exponential Utility Function

 When monetary value does not necessarily lead

to the most preferred decision, expressing the
value (or worth) of a consequence in terms of its
utility will permit the use of expected utility to
identify the most desirable decision alternative
 Utility is a measure of the total worth or relative
desirability of a particular outcome
 Reflects the decision maker’s attitude toward a
collection of factors such as profit, loss, and risk
62

31
 12/20/2016

 Example of a situation in which utility can

help in selecting the best decision alternative:
alternative
 Swofford Inc., a relatively small real estate
investment firm currently has two investment
opportunities that require approximately the
same cash outlay
 The cash requirements
q necessaryy p prohibit
Swofford from making more than one investment
at this time
 Consequently, three possible decision alternatives
may be considered
63

 The three decision TABLE 9.9: PAYOFF TABLE FOR

alternatives are: SWOFFORD INC.
SWOFFORD, INC
 d1 = make
investment A
 d2 = make
investment B
 d3 = do not invest
 The states of nature
are:
 s1 = real estate
prices
i go up
 s2 = real estate
prices remain
stable
 s3 = real estate
prices go down

64

32
 12/20/2016

 The best estimate of the probability that real estate prices

will go up P(s1) is 0.3;
 the best estimate of the probability that prices will remain
stable P(s2) is 0.5;
 the best estimate of the probability that prices will go down
P(s3) is 0.2
 The expected values for three decision alternatives are:

65

 Using expected value approach, the optimal decision is to

select
l investment A with h an expectedd value
l off $9,000.
 Is it really the best decision alternative?
 Consider some other relevant factors:
‐ Swofford’s current financial position is weak
‐ the firm’s president believes that, if the next investment
results in a substantial loss, Swofford’s future will be in
jeopardy
‐ even a loss of $30,000 could drive Swofford out of business
 Reasonable conclusion: the president would select d3,
believing that both investments A and B are too risky for
Swofford’s current financial position
66

33
 12/20/2016

 Resolve Swofford’s dilemma: determine Swofford’s

utility for the various outcomes
 The utility of any outcome is the total worth of that
outcome, taking into account all risks and
consequences involved.
 If the utilities for the various consequences are
assessed correctly,
correctly the decision alternative with the
highest expected utility is the most preferred.

67

Utility and Decision Analysis

 The procedure used to solve the Swofford
investment problem:
 Step 1: Develop a payoff table using monetary
values
 Step 2: Identify the best and worst payoff values
in the table and assign each a utility, with u(best
payoff)> u(worst payoff)

68

34
 12/20/2016

 Step 3. For every other monetary value m in the

original payoff table, do the following to
determine its utility:
a. Define the lottery such that there is a probability p of the
best payoff and a probability (1 ‐ p) of the worst payoff
b. Determine the value of p such that the decision maker is
indifferent between a g guaranteed p payoff
y of m and the
lottery defined in step 3(a)
c. Calculate the utility of m as follows:
U(M) = pU(best payoff) + (1 ‐ p)U(worst payoff)

69

 Step 4: Convert each monetary value in the

payoff table to a utility
 Step 5: Apply the expected utility approach to
the utility table developed in Step 4 and select
the decision alternative with the highest
expected utility

70

35
 12/20/2016

 We can compute
p the expected
p utility
y ((EU)) of
the utilities in a similar fashion as we
computed expected value

71

 Step 1: Develop a payoff table using monetary

values
l

72

36
 12/20/2016

 Step 2: Identify the best and worst payoff

values
l in the
h table
bl and d assign each
h a utility,
l
with u(best payoff)> u(worst payoff)
• Best payoff: $50.000
• Worst payoff: ‐ $50.000
• Utility of ‐$50.000
$50 000 = U(
U(‐50.000)
50 000) = 0
• Utility of $50.000 = U(50.000) = 10

73

 Step 3. For every other monetary value m in the

original payoff table, do the following to
determine its utility:
a. Define the lottery such that there is a probability p of the
best payoff and a probability (1 ‐ p) of the worst payoff
b. Determine the value of p such that the decision maker is
indifferent between a g guaranteed p payoff
y of m and the
lottery defined in step 3(a)
c. Calculate the utility of m as follows:
U(M) = pU(best payoff) + (1 ‐ p)U(worst payoff)

74

37
 12/20/2016

 Consider the process of establishing the utility of a

payoff
ff off $30,000
 First we ask Swofford’s president to state a
preference between a guaranteed $30,000 payoff
and an opportunity to engage in the following
lottery, for some probability of p that we select
 Lottery: Swofford obtains a payoff of $50,000
$50 000 with
probability p and a payoff of ‐$50,000 with
probability (1 ‐ p)

75

 If p is very close to 1, Swofford’s president would

prefer
f the h lottery
l to the
h guaranteed d payoff
ff off
$30,000 because the firm would virtually ensure
itself a payoff of $50,000
 If p is very close to 0, Swofford’s president would
clearly prefer the guarantee of $30,000
 as p increases continuously from 0 to 1, 1 the
preference for the guaranteed payoff of $30,000
decreases and at some point is equal to the
preference for the lottery

76

38
 12/20/2016

 At a specific value of p, Swofford’s president would have

equall preference
f f the
for th guaranteed
t d payoff
ff off $30,000
and the lottery; at greater values of p, Swofford’s
president would prefer the lottery to the guaranteed
$30,000 payoff
 Assume that when p = 0.95, Swofford’s president is
indifferent between the guaranteed payoff of $30,000
and the lottery. For this value of p, we can compute the
utility of a $30,000 payoff as follows:

77

 When p = 0.95, the expected value of the lottery is:

EV ((loterry)
y) = 0.95
95 (5
(50.000)) + 0.055 ((‐50.000)
5 )
= $45.000
 Swofford’s president is indifferent between the lottery (and
its associated risk) and a guaranteed payoff of $30,000. Thus,
Swofford’s president is taking a conservative, or risk‐
avoiding, viewpoint.
 A decision maker who would choose a guaranteed payoff
over a lottery with a superior expected payoff is a risk
avoider (or is said to be risk averse).
 The difference between the EV of $45,000 and the
guaranteed payoff of $30,000 is the risk premium that
Swofford’s president would be willing to pay to avoid the 5
percent chance of losing $50,000 78

39
 12/20/2016

TABLE 9.8: UTILITY OF MONETARY PAYOFFS

FOR SWOFFORD,
SWOFFORD INC
INC.

79

Step 4: Convert each monetary value in the payoff

table to a utility
TABLE 9.9: UTILITY TABLE FOR SWOFFORD,
INC.

80

40
 12/20/2016

Step 5: Apply the expected utility approach to the

utility table developed in Step 4 and select the
decision alternative with the highest expected
utility

81

Ranking of alternatives according to the president’s

y assignments
utility g and the associated monetary y values

Although investment A had the highest expected value of $9,000, the analysis
indicates that Swofford should decline this investment. The rationale behind is
that the 0.20 probability of a $50,000 loss was considered. The seriousness of this
risk and its associated impact on the company were not adequately reflected by
the expected value of investment A

82

41
 12/20/2016

Utility
y Functions
 Different decision makers may approach risk
in terms of their assessment of utility
 A risk taker is a decision maker who would
choose a lottery over a guaranteed payoff
when the expected value of the lottery is
inferior to the guaranteed payoff

83

 Analyze
y the decision p problem faced byy
Swofford from the point of view of a decision
maker who would be classified as a risk taker
 Compare the conservative point of view of
Swofford’s president (a risk avoider) with the
behavior of a decision maker who is a risk
taker

84

42
 12/20/2016

TABLE 9.10: REVISED UTILITIES FOR

SWOFFORD, INC., ASSUMING A RISK TAKER

85

TABLE 9
9.11: PAYOFF TABLE FOR SWOFFORD,, INC.

86

43
 12/20/2016

TABLE 9.12: UTILITY TABLE OF A

RISK TAKER FOR SWOFFORD,, INC.

 Using the state‐of‐nature probabilities P(s1) = 0.3, P(s2) = 0.5, and P(s3) = 0.2, the
expected utility for decision alternative is:
 EU(d2 ) = 0.3 (10) + 0.5 (1.5 ) + 0.2 (1.0 ) = 3.95
 EU(d1 ) = 3.50
 EU(d3 ) = 2.50
 The analysis recommends investment B, with the highest expected utility of 3.95 87

 Utility function for a risk avoider shows a

diminishing marginal return for money
 Utility function for a risk taker shows an
increasing marginal return
 These values can be plotted on a graph (Figure
9.11) as the utility function for money
 Top curve is utility function for risk avoider
 Bottom curve is utility function for risk taker
 Utility function for a decision maker neutral to risk
shows a constant return (middle line)

88

44
 12/20/2016

89

 The followingg characteristics are associated

with a risk‐neutral decision maker:
 The utility function can be drawn as a straight line
connecting the “best” and the “worst” points
 The expected utility approach and the expected
value approach applied to monetary payoffs result
in the same action

90

45
 12/20/2016

Exponential
p Utility
y Function
 Used as an alternative to assume that the
decision maker’s utility is defined when
decision maker provides enough indifference
values to create a utility function
 All the exponential utility functions indicate
that the decision maker is risk averse

91

92

46
 12/20/2016

 The R p parameter in equation

q (9.7)
(9 7) represents
p
the decision maker’s risk tolerance; it controls
the shape of the exponential utility function
 Larger R values create flatter exponential
functions, indicating that the decision maker
is less risk averse (closer to risk neutral)
 Smaller R values indicate that the decision
maker has less risk tolerance (is more risk
averse)
93

 Example: If the decision maker is comfortable

accepting a gamble with a 50 percent chance of
winning $2,000 and a 50 percent chance of losing
$1,000, but not with a gamble with a 50 percent
chance of winning $3,000 and a 50 percent chance
of losing $1,500, then we would use R = $2,000 in
equation
ti (9.7)
( )

94

47