UNIT IV DECISION THEORY

Introduction, Ingredients of Decision problems. Decision Making under

Uncertainty, Cost of Uncertainty – under risk, under perfect
information; Decision tree, Construction of Decision tree
• Decision Theory provides an analytical and systematic
approach to depict the expected result of a situation
when an alternative managerial actions and
outcomes are compared.
• Decision Theory is the combination of Descriptive and
Prescriptive business modelling approach to classify
the degree of knowledge like ignorance, uncertainty,
Risk, Certainty.
Decision alternatives:
There is a finite number of decision alternatives available with the
decision maker at each point in time when decision is made.
These alternatives are called courses of action or acts or
strategies
States of nature: These are the future conditions not under the
control of decision maker may be economy like inflation, political
development, weather condition etc.
• Payoff:
A numerical value resulting from each possible combination of
alternatives and states of nature is called “Payoff”.
General form of Payoff Matrix:
ALTERNATIVES
States of Probability S1 S2 S3 -- -- Sn
Nature

N1 P1 p11 p12 p1n

N2 P2 p21 p22 p2n

N3 P3

Nm Pm pm1 pm2 pmn

Product Type Alternative Demand(000 Rs)
D1 D2 D3
A 14 66 118

B 13 77 141

C 1 73 145
 DECISION MAKING UNDER UNCERTAINITY
There are several different criteria for decision making in this situation
1.Optimism (Maximax or Minimin) criterion
2.Pessimism (Maximin or Minimax) criterion
3.Equal probabilities (Laplace Criterion)
4.Coefficient of optimism (Hurwiez criterion)
5.Regret (Salvage ) criterion
1.A food products company is contemplating the introduction of a
revolutionary new product with new packaging or replace the existing
product at much higher price (S1) or a moderate change in the
composition of the existing product with a new packaging at small
increase in price (S2) or a small change in the composition of existing
product with negligible increase in price (S3).The three possible states of
nature are i) high increase in sales (N1) ii)No change in sales (N2)
iii) Decrease in sales (N3).
1.Which strategy should be chosen for i. Maximax or Minimin criterion
ii.Maximin criterion iii. Minimax Regret criterion iv. Laplace criterion

States of Nature Strategies

S1 S2 S3

N1 7,00,000 5,00,000 3,00,000

N2 3,00,000 4,50,000 3,00,000

N3 1,50,000 0 3,00.000
 i.Maximax Criterion
1. The decision maker ensures that he should not miss the opportunity to
achieve the largest possible profit (maximax) or lowest possible
cost(Minimin)
2. Locate the maximum (or minimum) payoff values corresponding to each
alternative
3. Select an alternative with the best anticipated payoff value
This criterion ,decision maker selects an alternative with largest (or
lowest) possible payoffs,it is called Optimistic decision criterion
i. Maximax or Minimin criterion

States of Nature Strategies

S1 S2 S3
N1 7,00,000 5,00,000 3,00,000

N2 3,00,000 4,50,000 3,00,00

N3 1,50,000 3,00,000 3,00.000

COLUMN MAXIMUM 7,00,000 5,00,000 3,00,000

The maximum of the column maxima is 700,000. Hence the company should adopt strategy
S1
 i.Minimax or Maximin or Pessimism Criterion
1. The decision maker ensures that he would earn no loss (or pay no
more) than the specified amount.
2. Locate the minimum payoff in case of loss or maximum payoff in case of
profit corresponding to each alternative
3. Select an alternative with the best anticipated payoff value maximum
for profit and minimum for loss or credit
This criterion ,decision maker is conservative about the future and always
anticipates worst possible outcome,it is called pessimistic decision
criterion
i. Maximin or Pessimistic or Walds criterion
States of Nature Strategies

S1 S2 S3

N1 7,00,000 5,00,000 3,00,000

N2 3,00,000 4,50,000 3,00,000

N3 1,50,000 0 3,00.000

COLUMN MINIMUM 1,50,000 0 3,00,000

The maximum of the column minima is 300,000. Hence the company should adopt strategy
S3.
3.Equal Probabilities/Laplace Method
Since the probabilities of states of nature are not known, it is assumed
that all states of nature will occur with equal probability.
1.Assign equal probability value to each state of nature by using the
formula 1/ states of nature.
2.Compute the expected payoff for each alternative by adding all the
payoffs and dividing by the number of possible states of nature
Probability of states of nature*(payoff values for the combination of
alternatives)
3.Select the best expected payoff value (maximun for profit minimum
for cost)
Since we do not know the probabilities or states of nature assume that
they are equal. Here each state of nature has probability 1/3 of
occurrence.
Strategy Expected Return
S1 (7,00,000+3,00,00+1,50,000)/3=3,83,333

S2 (5,00,000+4,50,000+0)/3=3,16,666.66

S3 (3,00,000+3,00,000+3,00,000)/3=3,00,000
Since the largest expected return is from strategy S1,the executive must
select strategy S1.
 Regret or Savage Criterion
This criterion is also known as ‘opportunity loss criterion’ or ‘Minimax
regret criterion’.
From the given payoff matrix, develop an opportunity loss matrix
1. Find the best payoff corresponding to each state of nature
2. Subtract all other entries in that row from this value.
3. For each course of action identify the worst or maximum regret value.
4. Select the course of action with the smallest anticipated opportunity
loss value.
States of Nature Strategies

S1 S2 S3

N1 7,00,000 5,00,000 3,00,000

N2 3,00,000 4,50,000 3,00,000

N3 1,50,000 0 3,00.000
 States of Nature Strategies

S1 S2 S3

N1 7,00,000- 7,00,000-5,00,000=2,00,000 7,00,000-

7,00,000=0 3,00,000=4,00,000

N2 4,50,000- 4,50,000-4,50,000=0 4,50,000-

3,00,000=1,50,000 3,00,000=1,50,000

N3 3,00,000- 3,00,000-0=3,00,000 3,00.000-

1,50,000=1,50,000 3,00,000=0

Column maximum 1,50,000 3,00,000 4,00,000

Minimax regret
Hence the company should adopt minimum opportunity loss strategy S1.
 HURWICZ CRITERION
Hurwicz approach suggests that decision maker selects an alternative that
maximizes H
H(criterion of realism)=α(max. in column)+ (1-α)(minimum in column)
1. Decide the Coefficient of Optimism α and then coefficient of
Pessimism (1- α)
2. For each alternative select the largest and lowest payoff value and
multiply these with α and (1- α) values respectively. The calculate the
weighted average H by using the above formula.
3. Select an alternative with the best anticipated weighted average payoff
value.
1.Which strategy should be chosen for i. Maximax or Minimin criterion
ii.Maximin criterion iii. Minimax Regret criterion iv. Laplace criterion

States of Nature Strategies

S1 S2 S3
N1 7,00,000 5,00,000 1,50,000

N2 3,00,000 4,50,000 0

N3 1,50,000 3,00,000 3,00.000

H FOR S1= 0.6* 7,00,000+(1-0.6)*1,50,000=4,80,000

H FOR S2= 0.6* 5,00,000+(1-0.6)*3,00,000=4,20,000
H FOR S3= 0.6* 3,00,000+(1-0.6)*0=1,80,000.Maximum H value is s1 strategy
Indicate the decision taken under following approaches i)Maximax criterion
ii)Maximin iii) Laplace iv) Regret/Savage v) Hurwicz criterions
STRATEGY STATES OF NATURE
N1 N2 N3 N4
S1 4000 -100 6000 18000

S2 20,000 5000 400 0

S3 20,000 15,000 -2000 1000
 S1 S2 S3

N1 4000 20,000 20,000

N2 -100 5000 15000

N3 6000 400 -2000

N4 18000 0 1000
DECISION MAKING UNDER RISK
• Decision making under risk is a probabilistic decision situation in
which m ore than one states of nature exists and the decision maker
has sufficient information to assign probability values to the likely
occurrence of each of these states.
• Expected Monetary value (EMV):
EMV for given course of action is the weighted sum of possible
payoffs for each alternative.It is obtained by summing payoffs for each
course of action multiplied by the probabilities associated with each
states of nature.
• EMV (course of action)= Σ pij* pi (i=1….m)
where m= number of possible states of nature
pi= probability of occurrence of states of nature
pij=payoff associated with states of nature Ni and course of action Sj
1.The manager of a flower shop promises its customer delivery within 4
hrs on all flower orders. All orders purchased on the previous day and
delivered to Parker by 8AM the next morning.The daily demand are as
follows.
The manager purchases roses for Rs.10 per dozen and sells them at
Rs.30.All unsold roses are donated to local hospital. How many dozens of
roses should Parker order each evening to maximize its profits.What is the
optimum expected profit?

Dozens of 70 80 90 100
roses
Probability 0.1 0.2 0.4 0.3
Sol: Marginal Profit =Selling price-Cost=3-1= Rs.2
Marginal loss=Loss on unsold roses=Rs.1

Conditional Profit= MP*Roses sold-ML*Roses not sold

= 2D if D>=S
=2D-1(S-D)= 3D-1S if D<S
If D=3,S=5
C.P= 3*2-(5-3)*1
=6-2=Rs. 4
States of Probabil Conditional Profit (payoff) due to course of Expected Payoff
nature(De ity action S
mand) 70 80 90 100 70 80 90 100

70 0.1 140 130 120 110(140- 14 13 12 11

30)
80 0.2 140 160 150 140 28 32 30 28

90 0.4 140 160 180 170 56 64 72 68

100 0.3 140 160 180 200(100*2) 42 48 54 60

Expected 140 157 168 168

Monetary
Value EMV
• Since the highest EMV is Rs .168 is corresponding course of action
90,the flower shop should purchase 9 dozens of roses everyday.
2.A retailer purchases cherries for Rs .50 and sells them at Rs.80 a case.Any
case unsold at the end of the day can be disposed of next day at salvage
value of Rs.20 per case.Find the no.of cases the retailer should purchase per
day to maximise his profit.

Cases sold 15 16 17 18
No.of 12 24 48 36
days(data for
120 days) 0.1 0.2 0.4 0.3

Sol.Marginal Profit =Rs.80-50 =Rs.30

Marginal loss= Rs.50-20=Rs.30
Conditional profit=30 D if D>=S
=30D-30(S-D) if D<S
Expected payoff and EMV table

States of Probabili Conditional Profits (S) Expected payoff

Nature( ty 15 16 17 18
D)

15 0.1 450 420 15*30- 360(15*30- 45 42 39 36

(450-30) 30*2=390 3*30)

16 0.2 450 480 450 420 90 96 90 84

17 0.4 450 480 510 480 180 192 204 192

18 0.3 450 480 510 540 135 144 153 162

EMV
3.The probability of the demand for lorries for hiring on any day in a given
district ;Lorries have fixed cost of Rs.90 and variable cost of Rs.200.
If lorry-hire company owns 4 lorries,what is the daily expectation? If the
cpy is about to go into business and currently has no lorries ,how many
lorries should it buy?
No.of 0 1 2 3 4
lorries
demanded
Probability 0.1 0.2 0.3 0.2 0.2
Sol:
No.of lorries 0 1 2 3 4
demanded
Payoff with 4 0-90*4=-360 200-90*4=- 400-90*4=40 600-90*4=360 800-90*4=440
lorries 160

The daily expectation (Expected payoff) is obtained by multiplying

payoffs with corresponding probabilities.
Daily expectation=-360*0.1+-160*0.2+40*0.3+240*0.2+440*0.2=Rs.80
Conditional Payoff values table

Demand Prob Conditional payoff

0 1 2 3 4
0 0.1 0 -90*1=-90 -90*2=-180 -90*3=-270 -90*4=-360
1 0.2 0 200-90=110 200-180=20 200-270=- 200-360=-
70 160
2 0.3 0 110 200*2- 400- 400-360=40
180=220 270=130
3 0.2 0 110 220 600- 600-
270=330 360=240
4 0.2 0 110 220 330 800-
360=440
Expected Payoff values & EMV table

Demand Prob Expected payoff

0 1 2 3 4

0 0.1 0 -9 -18 -27 -36

1 0.2 0 22 4 -14 -32

2 0.3 0 33 66 39 12

3 0.2 0 22 44 66 48

4 0.2 0 22 44 66 88

EMV 0 90 140 130 80

• Since EMV Rs.140 for the course of action 2 is higher,the company
should buy 2 lorries.
Expected Opportunity Loss (EOL)
• An alternative approach to maximizing EMV is EOL.It is also called as
Expected value of regret.

• EOL= Σ lij* pi
where lij=opportunity loss due to state of nature Ni and strategy Sj

• EOL is the amount of payoff that is lost by not selecting the course of action
that has the greatest payoff for the state of nature that actually occur. The
course of action due to which EOL is minimum is recommended.
1.A cpy manufactures goods for a market in which the technology of the
product is changing rapidly.The R &D has developed a new product which
requires Rs.60,000 for development testing.
The cpy has 100 customers and each customer might purchase at the most
one unit of the product. Market research suggests that S.P is Rs 6000 for
each unit and total variable costs is Rs.2000 for each unit.
The probability distribution related proportion of customers who will buy
the product as follows.
Determine the expected opportunity loss (EOL) and state whether the cpy
should or should not develop the product.
Proportion 0.04 0.08 .12 0.16 0.20
of
Customers
Probability 0.1 0.1 0.2 0.4 0.2
Sol:If p is the proportion of customers who purchases new product the
conditional profits= (6000-2000)*100p-60,000=Rs(4,00,000p-60,000)
Conditional Profit Values(Payoffs)
States of nature Conditional Profit(4,00,000p-60,000)
Course of action
S1(Develop) S2(Do not Develop)
0.04 -44,000 0

0.08 -28,000 0

0.12 -12,000 0

0.16 4000 0

0.2 20,000 0
States of Probability Conditional profit Opportunity Loss (Rs.)
Nature S1 (Develop) S2(do not S1 S2
Develop)

0.04 0.1 -44,000 0 44,000 0

0.08 0.1 -28,000 0 28,000 0

0.12 0.2 -12,000 0 12,000 0

0.16 0.4 4000 0 0 4000

0.20 0.2 20,000 0 0 20,000

EOL=Σ lij* pi 9600 5600

• Using the given estimates of probabilities associated with the states
of nature EOL for each course of action is given by
EOL(s1)=0.1*44,000+0.1*28,000+0.2*12,000+0.4*0+0.2*0=Rs.9600
EOL(S2)= 0.1*0+0.1*0+0.2*0+0.4*4000+0.2*20,000=Rs.5600

Since the cpy seeks to minimize the expected opportunity loss, the cpy
should select course of action S2(do not develop the product) with
minimum EOL.
Expected value of Perfect Information(EVPI)
• EVPI=Expected profit with perfect information(EPPI)-Expected profit
without perfect information
= Σ max(pij)* pi –EMV
Where pij=Best payoff when action s1 is taken in the presence of states
of Nature N1
Pi=Probability of States of Nature
EMV= Expected monetary value
1.A certain piece of equipment has to be purchased for construction project at a
remote location. This equipment consists of expensive part which is subject to
random failure. Their unit cost is Rs 1500 and they have no scrap value.
If the part fails on the job no spare part is available, the part will have to be
manufactured on special order basis. The total cost including down time is
estimated to be Rs 9000 for each occurrence.
Based on previous experience with similar parts, the following probability estimates
of the number of failures expected over the duration of the project are provided.
Failure 0 1 2
Probability 0.80 0.15 0.05

a) Determine optimal EMV and optimal number of spares to purchase initially.

b)Based on opportunity losses,determine the optimal course of action and
optimum value of EOL.
c)Determine Expected profit with perfect information and Expected value of perfect
information.
Sol: Let N1 (no failure) ,N2,N3(one,two failures) be the possible states of
nature. Similarly S1 (no spares purchased),S2,S3 (one,two spares purchased)
be possible courses of action or strategies.
State of Courses of Purchase Downtime Total
Nature action cost(Rs.) cost (Rs) conditional
cost

0 0 0 0 0
0 1 1500 0 1500
0 2 3000 0 3000
1 0 0 9000 9000
1 1 1500 0 1500
1 2 3000 0 3000
2 0 0 18000 18000
2 1 1500 9000 10,500
2 2 3000 0 3000
Expected Payoff and EMV
State of Probability Conditional cost due to course of action Weighted Cost due to course of action
Nature S1 (0) S2 (1) S3(2) S1 S2 S3
N1 0.8 0 1,500 3,000 0.8*0=0 1200 2400

N2 0.15 9,000 1,500 3000 9000*0.15= 225 450

1350
N3 0.05 18,000 10,500 3000 18000*0.05 525 150
=900

EMV 2250 1950 3000

Since the weighted cost Rs.1950 is lowest due to course of action S2 it
should be chosen.If the EMV is expressed in terms of profit then
EMV= EMV (S2)=-Rs 1950
b) To determine EOL
States of Conditional Cost due to strategies Conditional Opportunity Loss
Nature S1 S2 S3 S1 S2 S3

N1(No) 0 1500 3000 0 1500 3000

N2(1) 9000 1500 3000 7500 0 1500

N3(2) 18000 10500 3000 15000 7500 0

EOL
States of Probability Conditional Cost Weighted Cost due to course of action
Nature

N1 0.8 0 1500 3000 0.8*0=0 1200 2400

N2 0.15 7500 0 1500 0.15*7500=1125 0 225

N3 0.05 15000 7500 0 0.05*5000=750 375 0

EMV 1875 1575 2625

EOL=EOL(S2)=Rs 1575 adopt course of action S2 and purchase one spare.

c)EPPI can be obtained by selecting optimal course of action for each state
of nature multiplying by probabilities then summing these products.

States of nature Probability Optimal Cost of optimal course of action

course of Conditional Cost Weighted Opportunity
action loss
N1 0.80 S1 0 0.8*0=0

N2 0.15 S2 1500 0.15*1500=225

N3 0.05 S3 3000 0.05*3000=150

Total =375

Expected value of perfect information EVPI =EPPI-EMV*

=375-(-1950)=Rs. 1575
It os observed that EVPI=EOL*=Rs 1575
2. XYZ company manufactured parts for passenger cars and sells them in lots of
10,000 parts each. The company has a policy of inspecting each lot before it is
actually shipped to retailer.
Five inspection categories established for quality control, represent the
percentage of defective item contained in each lot. The daily inspection chart
for past 100 inspection shows 2 courses of action
a) S1: shut down entire plant and thoroughly inspect each machine
b) S2:Continue the production as it now exists but offer the customer a refund
for defective items discovered & returned.The first alternative will cost Rs
600 while the second alternative will cost cpy Rs.1 for each defective item
returned.
i) What is the optimum decision of the Cpy?
ii) Find the EVPI?
Rating Proportion of Defective item Frequency

Excellent 0.02 25

Good 0.05 30

Acceptable 0.10 20

Fair 0.15 20

Poor 0.2 5