MANAGEMENT FOR ENGINEERS

good idea of
incomplete information about available alternatives but has a
alternative.
the probability of outcomes for each
determine the
While making decisions under a state of risk, managers must available
probability associated with each alternative on the basis of the
information and his experience.
Decision-making under Certainty:
knows with
A condition of certainty exists when the decision-maker
reasonable certainty what the alternatives are, what conditions are
associated with each alternative, and the outcome of each alternative. Under
conditions of certainty, accurate, measurable, and reliable information on
which to base decisions is available.

The cause and effect relationships are known and the future is highly
predictable under conditions of certainty. Such conditions exist in case of
routine and repetitivedecisions concerning the day-to-day operations of the
business.

3.7 DECISION-MAKINGUNDER UNCERTAINTY

Most significant decisions made in today's complex environment are
formulated under astate of uncertainty. Conditions of uncertainty exist
when the future environment is unpredictable and everything is in a state of
flux. The decision-maker is not aware of all available alternatives, the risks
associated with each, and the consequences of each alterrnative or their
probabilities.
The manager does not possess complete infornmation about the alternatives
and whatever information is available, may not be completely reliable. In the
face of such uncertainty,managers need tomake certain assumptionsabout
the situation in order to provide a reasonable framework for decision
making. They have to depend upon their judgment and experience for
making decisions.
The following choices are available before the decision maker in situations
of uncertainty.

(a) Maximum Criterion

(b) Minimax Criterion
(c) Maximin Criterion
(d) Laplace Criterion (Criterion of equally likelihood)
(e) Hurwiczalpha Criterion (Criterion of Realism) (criterion of rationality)
3.7.1 The maximax decision criterion (Criterion of optimism)
The term 'maximax is an abbreviation of the phrase maximum of the
maxima. An adventurous and aggressive decision maker may choose the act
Module 3 Page 3.26
 MANAGEMENT FOR ENGINEERS

that would result in the maximum payoff possible. Suppose for each act
there are three pOSSible payofts corresponding to three states of nature as
matrix.
given in the followingdecision
Pay off table
States of nature Decisions
Aj Az A3
220 180 100
160 190 180
Sz 140 170 200
S3
Maximum payoffs under each decision A,, Az, A3 are respectively
A, Az A3
220'190'200
The maximum of these three maximums is 220 which relates to A,.
Consequently, according to the maximax criteria, the decision is to choose
A,.
3.7.2 The minimax decision criterion
Minimax is just opposite to maxima. Application of the minimax criterion
requires a table of losses instead of gains. The losses are the costs to be
incurred or the damage to be suffered for each of the alternative act and
states of nature. The minimax rule minimizes the maximum possible loss for
each course of action. The term 'minimax is an abbreviation of the phrase
minimum of the maxima. Under each of the various acts there is a maximum
loss and the act that is associated with the minimum of the various
maximum losses is the act to be undertaken according to the minimax
criterion. Suppose the loss table
States of nature Decisions
A Az A3
0 4 10
S1
3 6
S2 14
18
S3
Maximum losses incurred by the various decisions:
A, Az A3
18 14 10

dnd the minimum among these three maximums is 10 which is for act A3.
Thus, according to minimax criterion, the decision maker should take a
decision for A3.
Module 3 Page 3.27
 MANAGEMENT FOR ENGINEERS

of pessimism)
3.7.3 The maximindecision criterion (Criterion
called Waldian criterion
The ismaximin
This criterion of decision making stands for choice between
alternative courses of action assuming pessimistic view of nature. Taking
results in terms of pavoff a9
each act in turn, we note the worst possible
Suppose the nave
select the act which maximises the minimum pay off.
table is

States of nature Decisions

Aj Az A3
-80 -60 -20
S1
-30 -10 -2
S2 15 7
S3 30
75 80 25
S4

Minimum under each decision A,, Az, A3 are respectively

-80, -60, -20

The Act Az is to be undertaken according to this criterion because it is the

maximum among minimum.
3.7.4 Laplace criterion:
As the decision maker has no information about the probability of
occurrence of variousevents, the decision maker makes a simple assumption
that each probability is equally likely. The expected pay off is worked out on
the basis of these probabilities. The act having maximum expected pay off is
selected.
Act
Events A, Az A
E, 20 12 25
E2 25 15 30
30 20 22

Ans: We associate equal probability for each

pay off are the averages of pay offs
event say 3
.So the expected
under each act.
A, ’:3 (20 +25 + 30) == 25
3

A, ’ (12 + 15 + 20) = 47
3
3 = 15.67
Az ’: (25 + 30 + 22) = 7
3
= 25.67
3

Since Ahas maximum

expected payoff, A, is the optimal act.

Module 3
Page 3.28
 MANAGEMENT FOR ENGINEERS

3.7.5
Harwicz alphacriterion
ethod is acombination of maximin criterion and maximin criterion.
In this method, the decision maker's degree of optimism is represented by
the coefficient of optimism, a varies between 0 and 1. When a = 0, there
is total pessimism and when a= 1, there is total optimism.
We find D,, D2, D; etc. connected with all strategies where D; = aM; +
(1-a) n; and MË is the maximum payoff of i th strategy and m, is the
minimum payoff of '1 th strategy. The strategy with highest of D,, Dz, ..is
hosen. The decision maker will specify the value of a depending upon his
level of optimism.
Exanple:
Events Act
A Az A3
20 12 25
E1
25 15 30
E2
30 20 22
E3
Ans: Let a = 6
For A,,maximum pay off = 30 and Minimum pay off = 20
. D, = (0.6 X 30) + (1 0.6)20 = 26
Similarly, for A, maximum pay off = 20 and minimum pay off = 12

D, = (0.6 x 20) + (1-0.6)12 = 16.8

For A3, maximum pay off = 30 and minimum pay off = 22
D = (0.6 × 30) + (1-0.6022 = 26.8
Since D, is max, select the act A3.
Question 14:
Afood products company is planning the introduction of a revolutionary
new product with new packing to replace the existing product at much
higher price (S;) or amoderate change in the composition of the existing
product with a new packaging at a small increase in price(S2) or a small
change in the composition of the existing except the word, New' with a
negl1gible increase in the price (S). The three possible states of nature of
events are (i) high increase in sales (N,) (i1) no change in sales (,) (ii)
decrease in sales (N). The marketing department of the conmpany worked
out the pay offs in terms of yearly new profits for each of the strategies on
these events. This is represented in the following table (in next page).
Which strategy should the executive concerned choose on the basis of
(a) Maximin criterion (b) Maximax criterion (c) Minimax regret eriterion
(d) Laplace Criterion
Module Page 3.29
 MANAGEMENT FOR ENGINEERS

Pay offs
States of nature
Strategies N, N, N3
S, 700 300 150
S, 500 450
300 300 300

Solution:
Rewriting the pay off table (with strategies in columns)
Strategies
States of nature S S S3
N, 700 500 300
N2 300 450 300
150 0 300
N3
(a) Maximin criterion
Minimum pay offs for S, Sz, S3 are
S;
150 0 300

Max of these minima = 300 which relates to S

The executive should choose strategy Sa
(b) Maximax criterion
Maximum pay offs
S S S3
700 500 300

Maximum of these maxima = 700

The executive can choose Act S

(c) Minimax Regret Criterion

Opportunity Loss table
S S S.
N1 700 - 700 = 0 700 - 500 = 200 700-300 = 400
Nz 450- 300 = 150 450- 450 = 0 450 300 = 150
N3 300 - 150 = 150 300 -0 = 300 300 -300 = 0

Maximum opportunity loss for S,, Sz, S3

S; S3
150 300 400
Minimum of these maxima = 150 which relates to S;.
: The executive should choose strategy S, since it
minimizes the nmaximum
Module 3 Page 3.30
 MANAGEMENT FOR ENGINEERS
(d) Laplace criterion
Pay off table
S, Sz S3
700 500 300
300 450 300
150 0 300
Total 1150 950 900
Average (divided by 3) 383.3 316.67 300

Since the average is highest for strategy 1, the executive may select strategy
S
Question 15:
The research department of consumer products division has recommended
to the marketing department to launch a soap with three different perfumes.
The marketing manager has to decide the type of perfume to launch under
the following estimated pay off for the various levels of scales.
Estimate which type can be chosen under maximax, minimax, maximin,
laplace and Hurwicz Alpha criteria. (given a = 0.6)
Estimated levels of sales (units)
Types of perfumes 20,000 10,000 2000
250 15 10
II 40 20 5
III 60 25 3

Solution:
) Maximax criterion
Pay off table
Types of perfume
Levels of Sales II II
20000 250 40 60
10000 15 20 25
10 5 3
2000

Maximum for each type

Type I Type II Type III
250 40 60

Maximum of maximum = 250 which relates to type I

Select type l perfume

Module 3 Page 3.31

 MANAGEMENT FORENGINEERS

(ii) Minimax Criterion prepare Regret table

Regret table
Type
Sales
20,000 0 =
210 190
10,000 10 5
2,000 5 7
n
Maximum losses

Type I Typell Type lII

10 210 190

Minimum of these is 10 which relates to type I

Type l is preferred
(ii) Maximin Criterion
From the pay off table (Table I), Minimum pay off under each act
is

Type I Typell Type II

10 5 3

Maximum of these minima is 10 which relates to type I

Type Iis preferred.
(iv) Laplace Criterion
Pay off table
Type I Type II Type III
250 40 60
15 20 25
10 5 3
Total 275 65 88
Average 91.67 21.67 29.33

Average pay off is more for type I. So choose Type I perfume

(v) Hurwicz Alpha Criterion
Given a= 0.6,1-a= 1-0.6= 4

Max. pay off for type I=250, minimum pay off for type I= 10
D, =(0.6 × 250) + (0.4 x 10) = 150 + 4 = 154
Max. pay off for type I| =40, Minimum pay off for type Il =5
D, = (0.6 ×40) + (0.4 × 5) = 24 +2 = 26

Module 3 Page 3.32

 MANAGEMENT FOR ENGINEERS
Max. pay off for ty pe IIII = 60, Minimum pay off for type III =3
D. =(0.6 x 60)+ (0.4 X 3) = 36+ 1.2 = 37.2

D, is highest. So, select type I.

3.7.6 Modern Approaches to Decision-making under Uncertainty
ore are several modern techniques to improve the quality of decision
making under conditions of uncertainty.
The most important among these are:
(1)Risk analysis,
(2) Decision trees and
(3) Preference theory.
Risk Analysis:
Managers who follow this approach analyze the size and nature of the risk
involved in choosing a particular course of action.
For instance, while launching a new product, a manager has to carefully
analyze each of the following variables the cost of launching the product, its
production cost, the capital investment required,the price that can be set for
the product, the potential market size and what percent of the total market
it will represent.

Risk analysis involves quantitative and qualitative risk assessment, risk

management and risk communication and provides managers with a better
understanding of the risk and the benefits associated with a proposed course
of action. The decision represents a trade-off between the risks and the
benefits associated with a particular course of action under conditions of
uncertainty.
Decision Trees:
These are considered to be one of the best ways to analyze a decision. A
decision-tree approach involves a graphic representation of alternative
cOurses of action and the possible outcomes and risks associated with each
action.

Dy means of a "tree" diagram depicting the decision points, chance events

and probabilities involved in various courses of action, this technique of
decision-making allows the decision-maker to trace the optimum path or
course of action.
Preference or Utilityapproach
This is another Theory: to decision-making under conditions of
uncertainty. This approach is based on the notion that individual attitudes
towards risk vary. Some individuals are willing to take only smaller risks
("risk averters"), while others are willing to take greater risks ("gamblers").
Module 3 Page 3.33
 MANAGEMENT FOR ENGINEERS

Statistical probabilities associated with the

based on the assumption that decision-makersvarious courses of
will follow them action are
For instance, if there were a 60
might seem reasonable that a person percent chance of a decision being right i.
would take the risk. This may not ba
necessarily true as the individual might not
chances of the decision being wrong are 40 wish to take the risk, since tha
risk vary with events, with people and percent. The attitudes towards
positions.
Top-level managers usually take the
same managers who make a decisionlargest amount of risk.
that risks millions of However,
the
company in a given program with a75 percent chance of rupees of the
likely to do the same with their own success are not
money.
Moreover, amanager willing to take a 75 percent risk in one
not be willing to do so in another. situation may
Similarly, a top executive might launch an
advertising campaign having a 70 percent chance of success but
decide against investing in plant and machinery might
probability of success. unless it involves a higher
Though personal attitudes towards risk vary, two
things are certain.
Firstly,attitudes towards risk vary with
avertersin some situations and gamblerssituations,
in others.
i.e. some people are risk

Secondly, some people have a high aversion to risk, while others

aversion. have a low
Most managers prefer to be risk averters to a
also forego opportunities. When the stakes are certain extent, and may thus
be risk averters; when the stakes high, most managers tend to
are small, they tend to be gamblers.
3.8 DECISION MAKING UNDER RISK
In this situation the decision
maker has to face severalstates of nature. But
he has some knowledge or
experience which will
probability to the occurrence of each state of nature.enable him to assign
The objective is to
optimize the expected profit, or to minimize the opportunity loss.
For decision problem under risk, the most
popular methods used are EMV
(Expected MonetaryValue) criterion and EOL (Expected Opportunity Loss)
criterion.

3.8.1
Expected Monetary Value (EMV):
When the probabilities can be assigned to various states of
nature, it is
possible to calculate the expected pay off for each course of action. These
expected pay offs are known as EMV.

Module 3 Page 3.34

 MANAGEMENT FOR ENGINEERS
The conditional value of each event in the pay off table is
probabilityand the product is summed up.The multiplied by its
beact. The decision maker then resulting number is the EMV
act that leads to the optimum
selects from the available alternatives,
expected outcome
optimal EMV)..Optimal means maximization in the case of (that is the act with
nd minimization in the case of cost, expenditure etc. profit or income

ie EMV = (X, X p1) + (xz X P2)+. ... where x1, X,...

are pay offs and
P1,P2 ..are probabilities of states of nature

Consider the following example. Let the states of nature be S, and S, and the
alternative strategies be A, and Az. Let the pay off table be as shown below.
A Az
S, 30 20
35 30

Let the probabilities for the states of nature S, and S, be the respectively 0.6
and 0.4. Then,

EMV for A, = (30 x 0.6) + (35 x 0.4) = 18 + 14 = 32

EMV for A, = (20 x 0.6) + (30x 0.4)= 12 + 12 = 24
EMV for A, is greater. :.The decision maker willchoose the strategy A1.
The criterion of selecting the maximum expected pay off act, is some times
referred to as Baye's decision rule.
3.8.2 Expected Opportunity Loss (EOL):
When the probabilities for various states of nature are known, it is possible
tocalculate the expected loSses for each course of action.

EOL = (x, Xp;) + (x, Xp2)+....

Where x,,X2,..are losses and p,,p2,....are the probabilities of states of nature.
Ihese expected losses are known as Expected Opportunity Losses (EOL).
Under this criterion the strategy which has minimum Expected Opportunity
Loss (EOL) is chosen.
opportunity loss table.
Consider the following Example: Given below is an
1dnd Az are the strategies and S, and S, are
the states of nature.
Aj A,
0 10
S,
2 -5
S
Let the probabilities for twostates be 0.6 and 0.4
EOL for A, =(0 x 0.6) + (2x 0.4) = 0.8
=4
EOL for Az =(10 x 0.6) +(-5 x 0.4) =6-2

Module 3 Page 3.35

 MANAGEMENT FORENGINEERS

EOL for A, is least. Therefore, the strategy A, may be chosen.

Question 16:
Youare given the following pay - off matrix
States of
nature A1 A2 A3
S 25 -10 -125
S, 400 440 400
S 650 740 750

The probabilities of the states of nature are respectively 0 .1, 0.7 and 0.2.
Calculate and tabulate the EMV and conclude which of the acts can be
chosen as the best.

Solution:

Act A, Act A Act A

Prob x Pay off Prob x Pay off Prob x Pay off
25 x 0.1 = 2.5 -10× 0.1 = -1 -125 x 0.1 = -12.5
400 × 0.7 = 280 440x 0.7 = 308 440 x 0.7 = 280
650 x 0.2 = 130 740x 0.2 = 148 750 x 0.2 = 150
EMV = 412.5 EMV = 455 EMV = 417.5

Since EMV is maximum for A,, choose the Act A,

Question 17:

Find EMVfor all the A,, A,, Az acts from the table given below which is the
optimal act.
Events
Strategies
S Sz S3
(Acts) P(S) = 0.4,P(S,)
A1 2 8 15 = 0.3
Az -3 10 20 and P(S;) = 0.3
A3 -10 20 35

Solution:
Rewriting the pay off matrix to have strategies (acts) in columns
A, Az A3 Probability
S 2 -3 -10 0.4
S2 10 20 0.3
S3 15 20 35 0.3
Module 3 Page 3.36
 MANAGEMENT FOR ENGINEERS

Act A, Act Az Act A

Prob x Pay oft Prob x Pay off Prob x Pay off
2x 0.4 = 0.8 -3x 0.4 = -1.2 -10 x 0.4 = -4

8x 0.3 = 2.4 10 x 0.3 =3 20 x 0.3 = 6

20 x 0.3 = 6 35 × 0.3 = 10.5
15 x 0.3 = 4.5
EMV = 7.7 EMV = 7.8 EMV = 12.5

Highest EMV is A;. So A, is the optimal act.

Question 18:
Findwhich act is optimal, A or B. B
A
Pay off Probability Pay off Probability
0.5 10 0.4
12
0.2 20 0.3
15
20 0.3
20 0.3

Solution:
0.3) = 6+3 +6= 15
EMV for A= (12 >x 0.5)+ (15 x 0.2) + (20 x
x 0.3) = 4 +6+6= 16
EMV for B= (10 x 0.4) + (20 x 0.3) + (20
EMV is more for B. So Bis the optinmal act.
Question 19:
problem of choosing one of the products
Amanagement is faced with the matrix after market research for the two
tor manufacturing. The probability
products was as follows.
State of nature
Fair Poor
Act Good
0.15 0.10
ProductA' 0.75
0.10
0.60 0.30
Product 'B'
market
can make for different levels of
ne profit that the management
dcceptability of the products are as follows.
State of nature

Profit (in?) if market is

Poor
Good Fair
Act 5,000
15,000
Product 'A' 35,000 Loss of 3,000
20,000
Product "B 50,000

Module 3 Page 3.37

 ASA,S A,S AzS, AS, A,S,
A,
AzS, A,S,Calculation Call The units.Solution:The than past over costQuestion A20:
S, Preparing be
Since Solution:
management.
= = = = = = = = = alternative
them different experience small preferred Calculate
80 79 78 79 79 78 78 75 78 80 the of MANAGEMENT
ENGINEERS FOR
x x x x x5 x x x Cost
Selling Call bottles week 3ink the
A, , them per FairGood
5 5- 5- 5 5 5- 5 x of
- - - 5-78 expected
- pay price Az, states manufacturer of by expected
80 79 78 80 79 78 80 79 courses in bottle
-
Module 3 x x x x x3 x x x3 x -
price A3. S1, this the
end Pay the
3= 3 3 3 3 3 offs =3/- Sz, of weekly
3 of place. and and off
management. pay 15000
= = = = = = = =
of S3.
nature 5000 35000 value
400 395 390 395 395 390 390 390
390 ink action is
sells table off Pay
per 29000
EMV =
=?5/- (EMV) Product A
Page
3.38 - - - - - - - You soldproduces
demand x
X0.15 xoff
240 237 234 240 237 234 240 237 234 - bottle are at 0.10 0.75 of
are are during a X the
the price for =
= = = = = = = = = perproducing = = profit
160 158 156 155 158 156 150 153 156 demands required has a product 500 225026250 choice
bottle the certain
of
never
following ?5
to per type of
78 r been B
foformulate bottle. is -30002000050000 alternatives
units,
78 of greater, 35700
EMV = Pay
lessweek. ink Product
X B
79 units, The x x off
than 0.10 0.60
units pay
According at 0.30 X
in k a product
79 =
= = profit and
or off 78 total
units is -300 600030000
80 table. producedor advise
units. greater average should B
or to
80 the the
 MANAGEMENT FOR ENGINEERS
Pay - off Table
States of nature Act (Strategy)
A, A A
156 153 150
S2 156 158 155
S3 156 158 160

Note: We shallshow the states ot nature in

rows and Acts in columns.
Question 21:

Afactory produces 3varieties of

are given below.
fountain pens. The fixed and variable costs

Fixed cost Variable cost

Type 1 2,00,000 10
Type 2 73,20,000 78
Type 3 6,00,000 6

The likely demands under three

situations are given below:
Poor: 25,000, Moderate: 1,00,000, High: 1,50,000
If the price of each type is 20, prepare the
pay-off table after showing
necessary calculations.
Solution:
Let T,, Tz, T; stand for Type 1, Type 2, and Type 3. Let D, D,and D,
poor denmand, moderate demand and high demand stand for
respectively.
The pay-off (in thousands) = Sales revenue -Total variable cost -
Fixed cost.
hD, = (20 × 25) - (10 × 25) 200 = 500- 250 - 200 =
+50
l2D, = (20 × 25) (8 >x 25) - 320 = 500 200 - 320 =-20
13D, = (20 x 25) (6 x 25) 600 = 500- 150- 600 = -250
I1Dz = (20 x 100) (10 x 100) -200 = 2000 - 1000 - 200 =+800
2D = (20 × 100) (8x 100)-320 = 2000 800 -320 = +880
TD, = (20 x 100)) - (6 x 100))-600= 2000 600 600 =+800
T,D, =(20 × 150))- (10 x150) 200 =3000 -1500 200= +1300
T;D, =(20 × 150) (8 x 150) 320 =3000 -1200 320 =+1480
T,D, =(20 × 150) - (6 x150) 600 = 3000 900 600 =+1500

Module 3 Page 3.39

 MANAGEMENT FOR ENGINEERS

The pay off table (in 000s)

T; T2 T,

50 -20 -250
D1 880 800
800
D2 1480 1500
D3 1300

Question 22:
Anewspaper boy has the following probability of selling a magazine.
No. of copies sold Probability
10 0.1
11 0.3
12 0.4
13 0.2

Cost of acopy is 3And sale price is5. He can return magazine but each
for one Rupee only. Prepare pay off table. How many copies should he
order? Also find expected number of sales.
Solution:

We can apply either EMV criterion or EOL criterion. Let us apply EMV
criterion for which we have to calculate pay off. Number of copies to be
ordered are different courses of action. The copies ordered are 10, 11, 12, 13.
Denote them by A,,Az, Az, A4:
Similarly, number of copies demanded are 10, 11, 12 or 13. These demands
may be denoted by D,,D2, D3, D,. These are events, the pay off values are
calculated below:

A, D, = (10x 5) (10 x 3) = 20 A, D, = (10x 5) - (10 x 3) = 20

A,D, = (10 x 5) (11 x 3) + (1 x 1) = 18 A,D, = (11 x 5) - (11 x 3) = 22
A,D, = (10x 5) - (12 x 3) + (2 x 1) = 16 A,D, = (12 x 5) - (12 x 3) = 24
A,D, = (10 x 5) - (13 x 3) + (3 × 1) = 14 A,D, = (12 x 5) -(13 x 3)
+(1x 1) =22
A, D, = (10 x 5) - (10 x 3) = 20 A, D, = (10 x 5) (10 x 3) = 20
A,D, = (11 x5) - (11 x3) = 22 A,D, = (11 x5) -(11 x 3)= 22
A,D, = (11 x 5) - (12 x 3) + (1 x 1) = 20 A,D, = (12 x 5) - (12 x 3) = 24
A,D, = (11 x 5) - (13 x 3)+ (2 x 1) = 18 A,D, = (13 x 5) -(13 x 3)= 26

Module 3 Page 3.40

 MANAGEMENT FOR ENGINEERS
Pay of table
A, Az A3 A,
20 18
Di 16 14
20 22
D2 20 18
D3 20 22 24 22
D4 20 22 24 26

Given probabilities are 0.1, 0.3, 0.4, 0.2

Calculation of EMV for all the Acts

A1 A2 Az A4
20x 0.1 = 20 18 x 0.1=1.8 16 x 0.1 = 1.6 14 ×0.1 = 1.4
20× 0.3 = 60 22 × 0.3 = 6.6 20 x 0.3 = 6.0 18 x 0.3 = 5.4
20x 0.4 = 80 22 x 0.4 = 8.8 24 x 0.4 = 9.6 22 x 0.4 = 8.8
20 x 0.2 = 40 22 x 0.2 = 4.4 24 x 0.2 = 4.8 26 x 0.2 = 5.2
EMV = 20 EMV= 21.6 EMV = 22 EMV = 20.8

EMV for A, is greater and therefore A, is the optimal act.

No. of copies to be ordered= 12
To find the expected number of sale, we multiply number of copies
sold and probability.
Expected number of sales
= (10x 0.1) + (11 x 0.3) + (12 x 0.4) + (13 x0.2) = 11.7 = 12

Choosing optimal Act by EOL Criterion

Question 23:
Given below is a regret table
Events(state of nature)
Act
E3
250
A, 350
300 0
Az 0
100 100
A3 450

table are
DuPpose that the probabilities of the events in this
= 0.25
P(E) = 0.30: P(E,) = 0.45; P(E;)
Calculate the expected Joss of eachaction. Interpret.
Module 3 Page 3.41
 MANAGEMENT FOR ENGINEERS

Solution:
Rewrite the problem by taking Acts in columns and events in rows.
Opportunity loss table
(Regret table)
A, A, Az
E; 350 450
Ez 300 100
E3 250 0 100

Calculation of Expected Loss (EOL)

A A, A,
Loss x Prob Loss × Prob Loss x Prob
350 x 0.30 = 105 0x 0.30 = 0 450 x 0.3 = 135
0x 0.45 = 0 300 × 0.45 = 135 100 x 0.45 = 45
250 x 0.25 = 62.5 0x 0.25 = 0 100 x 0.25 = 25
EOL = 167.5 EOL = 135 EOL = 205

Least EOL is for A,.So A, is the optimal act.

Preparing Regret Table

Question 24:
A small ink manufacturer produces a certain tvpe of ink at total
average cost
of?3per bottle and sells at a price of ? 5 per bottle. The ink is produced over
the week-end and is sold during the following week. According to the past
experience the weekly demand has never been less than 78 or greater than
80 bottles.

You are required to formulate the loss table.

Solution:

Calculation of opportunity loss

A,S, = 0 (since production and sales are of equal quantity say 78)
AzS, = 1x 3=3 (since one unit of production is in excess whose cost =5)
AzS, = 2x3= 6 (since 2 units of production are in excess whose unit
at 3) cost 1s

A,S, = 1 x 2 =2 (since the demand of one unit is more than

profit for one unit is 2)
produced. The

Module 3 Page 3.42

 MANAGEMENT FOR ENGINEERS
similarlv A,S, = 0|since units of production = units of demand|
= 3, A, S, = 2x2 = 4, A,S, = 2 x 1 = 2
A;S, = 1x3 and A,S, = 0
Opportunity Loss table (Regret Table)
States of nature A, Az A3
3
2
S2 3
S3 4 2

Preparing Regret Table from pay off table

Note: When each element in the pay-off table stands for the gain (income or
nrofit) to get Regret table from it, subtract all the elements of each row from
the highest pay off that row. (Every row in a pay-off table representing a
state of nature)
Question 25:
Given below is a pay-off table. Prepare Regret table from it.
Acts
Events A1 Az A3
B1 4 -2 -10
B2 7 8 3

B3 12 25 48

Solution:
Regret Table
A1 Az Az Aj Az A3
B 44 4 -2 4--10 6 14
B2 8-7 8-8 8-3 1 5
B3 48 - 12 48 - 25 48 48 36 23 0

Expected Valueof Perfect Information(EVPI):

get perfect value with perfect information, select the highest pay off
írom every row and with these pay offs, find the expected value, i.e.,
Expected value with perfect information =m,p; +mzpzt... where
f, m2, ... are respectively the maximum pay offs of each row.

P,P, ... are the probabilities of corresponding states of nature.

(2) EMV of the
optimal act is the Maximum along the EMV's of all acts.

Module.3 Page 343

 MANAGEMENT FORENGINEERS

(3) Expeted value of perfect information (EVPI) is the difference between

the tw0.

1.e., EVPI=Expected value with perfect information - Optimum EMy

EVPl gives an upper bound of the amount which the decision
maker can spend for obtaining perfect information.
Question 26:
Compute EVPI from the Pay-off table given below known that P(S,) =
0.5, P(S,)=4 and P(S;) = 0.1
Pay off table
State of nature A1 Az A3
S 30 25 22
Sz 20 35 20
S3 40 30 35
Solution:

Aj Az A3
Pay off X prob Pay off x prob Pay off >X prob
30 x 0.5 = 15 25 x 0.5 = 12.5 22 x 0.5 = 11
20 x 0.4 = 8 35 × 0.4 = 14 20 x 0.4 = 8
40 × 0.1 = 4 30 × 0.1 =3 35 x 0.1 = 3.5
EMV = 27 EMV = 29.5 EMV = 22.5

The highest EMV is for the strategy A, and it is 29.5

Calculating expected value with perfect information, considering

highest profit in every now
Events Max. Pay offof
each row Probability Expected value
S.
(Profit x prob. )
30 0.5 15
S 35 0.4 14
S 40 0.1 4

:Expected pay off with perfect

information = 15 +14 + 4 =33
Thus the expected value of perfect information =
with perfect information - minimum EMV (EVPI) = Expected Value
=33 -29.5 =3.5
Question 27:
A grocery store with a bakery department is faced
many cakes to buy in order to meet the day's with the problem of how
demand. The grocer preters
Module 3 Page 3.44
 MANAGEMENT FOR ENGINEERS
ottosell day - old goOds in competition with fresh products. Leftover cakes
theretore, a complete loss. On the other hand, if a customer desires a
ae n all of them have been sold, the disappointed customer will buy
clewhere and the sales will be lost. The grocer has, therefore collected
intormation on the past sales or a selected 100-day period as shown in the
table below:

Sales per day No. of days Probability

25 10 0.1
26 30 0.3
27 50 0.5
28 10 0.1
100 1.0

Construct the payoff table. What is the optimal number of cakes that should
be bought each day? Find and interpret EVPI. A cake costs Rs. 0.80 and sells
for Rs. 1.

Solution:
Let A,,Az, Ag, A, stand for strategies and S, Sz, S3, S, stand for states of
nature.

Then A,, Az, Az, A4 respectively stand for stocking 25, 26, 27, 28 cakes and
S;, Sz, S3, S, respectively stand for demands for 25, 26, 27, 28 cakes.

Conditional payoff values can be obtained as explained in previous problem.

The payoff values thus obtained are given below.
Pay off table
Alternative Strategies
State of
A, (25) A, (26) As(27) A,(28) Prob.
nature
S, (25) 5.00 4.20 3.40 2.60 0.1
S, (26) 5.00 5.20 4.40 3.60 0.3
S,(27) 5.00 5.20 5.40 4.60 0.5
0.1
S,(28) 5.00 5.20 5.40 5.60
EMV 5.1 4.9 4.2

Students may obtain EMV values for all acts.

Max EMV is for Act A,which is equal to 5.10
woukt
us according to the EMV decision criterion, the store
stock 26 cakes.
Module 3 Page 3 45
 MANAGEMENT FOR ENGINEERS

For calculating Expected value with perfect

Highest pay off for each state of nature (i.e., each row) and
value.
Pay off Prob.
innformfindation,expectseleedct
Pay offx Prob.
5.00 0.10
5.20 0.30
0.50
0.50
1.56
5.40 2.70
5.60 0,10
0.56
5.32
.Expected values with perfect information =5.32
Expected value of perfect information (EVPI) = Expected Value
Information - Highest EMV = 5.32 5.10= 0.22 with Perfect
Note: In the case of Cost matrix, optimal strategy is that having
EVPI= Minimum EMV - Expected value with perfect least EMV
information. In the
case of cost matrix, expected value with perfect information is calculated
by selecting the least pay off of each row.
Question 28:
Given below is a pay-off table

Events Act
A1 A2 A3
E 5 -1 -17
E2 18 28 30
E3 25 50 80

What will be the optimal decision if the criterion followed is (i) maximax ")
minimax (ii) maximin (iv) Laplace (v) EMV(vi) EOL. Given probabilites10
various events are .3, 0.4, 0.3
respectively.
Solution:
Pay off table
A Regret table
A A; A Probability
E, 5 -1 Az A, P(E) = 0.3
-17
Ez 18 28
E1 6 22
P(E)= 0.4
26
E3 25 50 80
Ez 10 0 2 P(E)= 0.3
E 55 30 0
(i)
Maximax creation
Maximum pay off for the acts (from pay off table)
Module 3 Page 3.46
 MANAGEMENT FOR ENGINEERS

A1 Az A3
25 50 80
Maximum of these maxima = 80, (which relates to A).
: Optimal Act is A3
Minimax criterion
(i1)
Maximum loss for the Acts (From regret table)
A, Az A3
55 30 22
Minimum of these maxima = 22 which relate to A.
::Optimal Act is A3

(ii) Maximin criterion

Az A3
5 -17

Maximum of these minima =5, which relates to A1

.Optimal Act is Aj
(iv) Laplace criterion
Average of pay offs of the Acts
1 48
A, , (5 + 18 + 25) = -= 16

1 77
A,’(-1+ 28 +50) = 3
= 25.67
1 89
= 29.67
Az>;(-17 + 26+ 80) = 3
Highest average is 29.67, which relates to As.
:: The optimal Act is A3
(v) EMV creation
EMV for various acts (pay offxprob.)
A, >(5 x 0.3) + (18 x 0.4) + (25 x 0.3) = 1.5 + 7.2 + 7.5 = 16.2
A, >(-1 x 0.3) + (28 x 0.4) + (50 x 0.3) = -0.3 + 11.2 + 15
= 25.9

A,-17 × 0.3)+ (26 x 0.4) + (80 x 0.3)= -5.l + 10.4 + 24

= 29.3

Highest EMV is 29.3, which relates to Ai

·.A, is the optimal act

MMule i Page 3 47
 MANAGEMENT FORENGINEERS
(vi) EOL criterion

EOL for various acts are (Regretxprob.)

A, >0x 0.3) + (10 x 0.4) + (55 x 0.3) = 0+4 + 16. = 20.5
A, >6 x 0.3) + (0 x 0.4) + (30 x 0.3) = 1.8 +0+9 = 10.8
A, (22x0.3) + (2 x0.4) + (0 x 0.3) = 6.6 + 0.8 + 0 =7.4
Minimum EOL is 7.4, which relates to A3
: As is the optimal act
Bayesian Rule (Posterior analysis)
Bayesian rule of decision theory is an approach in which the
selects a course of action on rational basis decision maker
by using subjective evaluation of
probability based on experience, past performance, judgement etc.
For making use of the Baye's
problem, the decision maker has toPrinciple
assign
in the statistical decision
probabilities to each state of
nature. These probabilities represent the strength
belief. ie., a subjective evaluation of the decision maker's
regarding the likelihood of the
OCcurrence of the various states of nature.
After determining the probabilities, the
phase wise. The three phases are () priorBaye's principle must be used
analysis (ii) posterior analysis. analysis (ii) preposterior
A decision maker assigns
subjective evaluation of
probabilities to various events which is his
likelihood of
states on the basis of experience of past the occurrence of the various
probabilities are performance. When these prior
used, the procedure is known as prior
analysis.
If prior analysis reveals a high
EVPI,
obtained. Prior probabilities may then, additional information is to be
be revised on the basis of these
additional information. By applying Baye's theorem of
revised probabilities are computed. probability, the
These-probabilities
posterior probabilities. A further analysis of the are known as
problem using these
posterior probabilities give new expected pay offs. This revised
of the problem is known as
posterior analysis
analysis.
Preposterior analysis is done to assess the
information as against the expected value ofexpected value of sample
perfect information even
before selecting a sample for additional information. This analysis
involves the revision of probabilities using Baye's
analysis involves arriving at a decision after revising theorem. Posterior
probabilities.
Module 3 Page 3.48
 MANAGEMENT FOR ENGINEERS

DECISION TREES
3.9
instances, the choice of the best act is not made in one stage, and
In many probleminvolves a sequence of acts, events, acts, events. There
thedecisionnumber of basic alternatives, each leading to one of a number of
may be a
situations dependig onthe outcome of a certain random process. At each
number of other alternatives may be available which also
such1 situation, a
set of situations depending on another set of events and so on,
Jead to a new events. Discrete decision
with acts followed by events, followed by acts,
using decision tress. It
theorv problems can be represented pictorially
as they unfolds. In
chronically portrays the sequence of actions and eventsthat can be taken by
set of actions
below figure, square symbol precedes the the
decision maker. The round nodeprecedes set of events or states of nature
made. The nodes are connected
that could be encountered after decisions is
by branches.
possibilities or outcomes
Adecision tree isthe graphical depiction of all the
opportunity. It is a useful
to solve a specific issue or avail a potential all the probable
financial tool which visually facilitates the classification of
results in a given situation.
E1

2
A1
E2

E1
A2
3

E2

Analysis of Decision Trees

from right to left. The aim of
niter the tree has been drawn, it is scrutinized maker that means
alalysis is to determine the best strategy of the decision
a decision tree, managers
Pamal sequence of the decisions. To analyze
ast know adecision criterion, probabilities that are assigned to each event,
the chance events
revenues and costs for the decision alternatives and
that ocCur.
and costs in a decision
There are two possibilities to how toinclude revenues nodes where they
tree. One possibility is to assign them only to terminating with
are included in tthe conditional value of the decision criterion associated
the the first part of the tree to the
decisionsand events along the path from
Module J Page 3.49
 MANAGEMENT FOR ENGINEERS

end. However, it an be appropriate to assign revenues and costs

branches. This reduces the required arithmetic for calculating the valuest
the decision criterion for terminating nodes and focuses attention on
parameters for sensitivity analysis.
When analyzing a decision tree, managers must start at the end of the tree
and work backwards. They perform two kinds of calculations.
For chance event nodes managers calculate certainty equivalents related to
the events emanating from these nodes. Under the assumption that the
decision maker has a neutral attitude toward risk, certainty equivalent of
uncertain outcomes can be replaced by their expected value. At decision
nodes, the alternative with the best expected value of the decision criterion
is selected.

Steps in Decision Tree Analysis

1. The first step is understanding and specifying the problem area for
which decision making is required.
2. The second step is interpreting and chalking out all possible
solutions to the particular issue as well as their consequences.
3. The third step is presenting the variables on a decision tree along
with its respective probability values.
4. The fourth step is finding out the outcomnes of all the variables and
specifying it in the decision tree.
5. The last step is highly crucial and backs the overall analysis of this
process. It involves calculating the EMV values for all the chance
nodes or options, to figure out the solution which provides the
highest expected value.
Benefits of Decision Trees:
Depicts Most Suitable Project/Solution: It is an effective means of picking
out the most appropriate project or solution after examining all the
possibilities.
Easy Data Interpretation and Classification: Not being rocket science,
decision tree eases out the process of segregation of the acquired data into
different classes.

Assist Multiple Decision-Making Tools: It also benefits the decision-maker

by providing input for other analyticalmethods like nature's tree.
Considers Both, Categorial and Numerical Data: This technique takes into
consideration the quantitative as well as the qualitative variables for better
results.

Module 3 Page 3.50