Decision Analysis

Five Steps in Decision Making

1. Clearly define the problem
2. List all possible alternatives or courses of action
-Decision variable
3. Identify all possible outcomes for a decision problem
-Chance variable/states of nature
A particular event that is relevant to the decision but is
beyond our control.
4. Identify the payoff for each alternative & outcome
combination
-Payoff function
5. Use a decision modeling technique to choose an
alternative
 Caselet

A bookstore sells a particular book of tax laws for Rs 100. It

purchases the book for Rs 80 per copy. Since some of the tax
laws change every year, the copies unsold at the end of a year
become outdated and can be disposed off for Rs 30 each.
According to past experience, the annual demand for this
book is between 18 and 23 copies. Assuming that the order
for this book can be placed only once during the year, the
problem before stores manager is to decide how many copies
of the book should be purchased for the next year.
Demand between 18 and 23 copies, six possible events:
E1: 18 copies are demanded E2: 19 copies are demanded
E3: 20 copies are demanded E4: 21 copies are demanded
E5: 22 copies are demanded E6: 23 copies are demanded

Six possible strategies or courses of action:

A1: 18 copies are ordered A2: 19 copies are ordered
A3: 20 copies are ordered A4: 21 copies are ordered
A5: 22 copies are ordered A6: 23 copies are ordered
 Payoff Function

Let D be demand in units for book

Q denote quantity to be purchased (course of action)
P be the profit

When D>=Q P=20Q

When D<Q P=20D-50(Q-D)
 Pay off Table

Act, Aj
Event Ei A1:18 A2:19 A3:20 A4: 21 A5: 22 A6: 23
E1: 18 360 310 260 210 160 110
E2: 19 360 380 330 280 230 180
E3: 20 360 380 400 350 300 250
E4: 21 360 380 400 420 370 320
E5: 22 360 380 400 420 440 390
E6: 23 360 380 400 420 440 460
 Types of Decision-Making Environments

Type 1 Decision making under certainty

Type2 Decision making under uncertainty

Type 3 Decision making under risk

Type 4 Decision making in competitive environment

Decision making under
uncertainty
 Decision Making Under Uncertainty
Probabilities of the possible outcomes are not known

How to decide which decision alternative to select ?

Several Ways to decide

Depends on the outlook of decision maker
No universal right decision
Pessimistic vs. Optimistic vs. Rationalistic
Several criteria for decision making are possible
Several strategies are available for decision making
 Decision Making Under Uncertainty

Rule 1: Pessimism (Maximin or Minimax) Criterion

Rule 2: Optimism (Maximax or Minimin) Criterion

Rule 3: Criterion of Realism or Hurwicz Criterion

Rule 4: Equally Likely Criterion or Laplace Criterion

Rule 5: Regret (Savage) Criterion or Minimax

Regret Criterion
 Rule 1: The Maximin payoff criterion
Adopted by pessimistic decision-makers.
Picks the decision alternative whose minimum
payoff gives the maximum among all.
Act, Aj
Event Ei A1:18 A2:19 A3:20 A4: 21 A5: 22 A6: 23
E1: 18 360 310 260 210 160 110
E2: 19 360 380 330 280 230 180
E3: 20 360 380 400 350 300 250
E4: 21 360 380 400 420 370 320
E5: 22 360 380 400 420 440 390
E6: 23 360 380 400 420 440 460
Min 360 310 260 210 160 110
Maxmin
Note: Minmax criterion

When dealing with costs

Maximum cost associated with each alternative considered

The alternative which minimizes this maximum cost is chosen.

Best of the worst

 Rule 2: The Maximax payoff criterion
Decision alternative which maximizes the maximum return
Optimistic Approach

Act, Aj
Event Ei A1:18 A2:19 A3:20 A4: 21 A5: 22 A6: 23
E1: 18 360 310 260 210 160 110
E2: 19 360 380 330 280 230 180
E3: 20 360 380 400 350 300 250
E4: 21 360 380 400 420 370 320
E5: 22 360 380 400 420 440 390
E6: 23 360 380 400 420 440 460
Max 360 380 400 420 440 460
Note: The Minimin criterion

When dealing with costs

Minimum cost associated with each alternative is considered

Alternative which minimizes the minimum cost is chosen.

 Rule 3: Criterion of Realism or Hurwicz Criterion
Decision makers view may fall between the extremes
posed by the optimist and pessimist criteria

Use the coefficient of realism () to estimate the decision

makers optimism, 0 < < 1

Realism payoff for alternative =

x (max payoff for alternative) + (1- ) x (min payoff for
alternative)

Choose the alternative with Maximum Realism payoff

 Let =.6
Realism payoff = x (max payoff for alternative) + (1- ) x
(min payoff for alternative)

Act, Aj
Event Ei A1:18 A2:19 A3:20 A4: 21 A5: 22 A6: 23
E1: 18 360 310 260 210 160 110
E2: 19 360 380 330 280 230 180
E3: 20 360 380 400 350 300 250
E4: 21 360 380 400 420 370 320
E5: 22 360 380 400 420 440 390
E6: 23 360 380 400 420 440 460
Max 360 380 400 420 440 460
Min 360 310 260 210 160 110
Realism 360 352 344 336 328 320
payoff
Note: In case of costs

Minimum cost for each course of action is multiplied by

Maximum cost for each alternative is multiplied by 1-

The sum of the products for each action strategy is obtained

The alternative for which the sum is the least is selected.

 Rule 4: Equally Likely Criterion or Laplace Criterion
Assume all outcomes are equally likely
Choose the alternative which maximizes the average
payoff (minimum for costs)
Act, Aj
Event Ei A1:18 A2:19 A3:20 A4: 21 A5: 22 A6: 23
E1: 18 360 310 260 210 160 110
E2: 19 360 380 330 280 230 180
E3: 20 360 380 400 350 300 250
E4: 21 360 380 400 420 370 320
E5: 22 360 380 400 420 440 390
E6: 23 360 380 400 420 440 460
Expected 360 368.3 365 350 323.3 285
payoff
 Rule 5: Regret (Savage) Criterion or Minimax
Regret Criterion

Obtain the regret matrix from the pay-off matrix

The maximum regret value corresponding to each of the

strategies is determined.

Strategy/Decision alternative which minimizes the

maximum regret is chosen
 Opportunity Loss or Regret Table
Opportunity loss- Amount of pay off foregone by not adopting
the optimal course of action (which would give the highest
payoff for each possible event).
Act, Aj
Event Ei A1:18 A2:19 A3:20 A4: 21 A5: 22 A6: 23
E1: 18 0 50 100 150 200 250
E2: 19 20 0 50 100 150 200
E3: 20 40 20 0 50 100 150
E4: 21 60 40 20 0 50 100
E5: 22 80 60 40 20 0 50
E6: 23 100 80 60 40 20 0

Hint: Payoff matrix can be transformed into opportunity loss matrix by

subtracting from the highest profit value in each row, all other values in
that row.
 Regret Table

Act, Aj
Event Ei A1:18 A2:19 A3:20 A4: 21 A5: 22 A6: 23

E1: 18 0 50 100 150 200 250

E2: 19 20 0 50 100 150 200
E3: 20 40 20 0 50 100 150
E4: 21 60 40 20 0 50 100
E5: 22 80 60 40 20 0 50
E6: 23 100 80 60 40 20 0

Max 100 80 100 150 200 250

 Decision Making under Risk
(with Probabilities)
Decision maker has sufficient information to assign
probability to the likely occurrence of each outcome
(state of nature).

Maximum Likelihood Criterion

Expectation Principle
 The Maximum Likelihood Criterion
Decision alternative with maximum payoff for the most
likely state of nature.
Act, Aj
Event Ei Prob A1:18 A2:19 A3:20 A4: 21 A5: 22 A6: 23
E1: 18 .05 360 310 260 210 160 110
E2: 19 .10 360 380 330 280 230 180
E3: 20 .30 360 380 400 350 300 250
E4: 21 .40 360 380 400 420 370 320
E5: 22 .10 360 380 400 420 440 390
E6: 23 .05 360 380 400 420 440 460
Most likely state of nature = E4 (highest prob), demand for 21 books
Maximum payoff in case of E4 = 420, thus order 21 books
Reasonable when probability of a particular event is
predominantly larger than the probabilities of others.
 Expectation Principle

EP/EMV- Expected Payoff/ Expected Monetary Value

ER/EOL Expected Regret/ Expected Opportunity Loss

 Expected Monetary Value- EMV

Expected payoff or Expected Monetary Value (EP/EMV) for

each strategy is calculated

Expected Monetary Value (EMV) uses the probabilities to

calculate the average payoff for each alternative

EMV (for alternative i) =

(probability of outcome) x (payoff of outcome)
 Expected Monetory Value/Pay off

Act, Aj
Event Ei Prob A1:18 A2:19 A3:20 A4: 21 A5: 22 A6: 23
E1: 18 .05 360 310 260 210 160 110
E2: 19 .10 360 380 330 280 230 180
E3: 20 .30 360 380 400 350 300 250
E4: 21 .40 360 380 400 420 370 320
E5: 22 .10 360 380 400 420 440 390
E6: 23 .05 360 380 400 420 440 460
EMV 360 376.5 386 374.5 335 288.5

Pick the strategy corresponding to maximum expected payoff.

i.e., buy 20 copies
 Expected Opportunity Loss (EOL)

How much regret do we expect based on the probabilities

EOL (for alternative i) =

(probability of outcome) x (regret of outcome)
 Expected Opportunity Loss or Expected Regret (EOL/ER)
Act, Aj
Event Ei prob A1:18 A2:19 A3:20 A4: 21 A5: 22 A6: 23
E1: 18 .05 0 50 100 150 200 250
E2: 19 .10 20 0 50 100 150 200
E3: 20 .30 40 20 0 50 100 150
E4: 21 .40 60 40 20 0 50 100
E5: 22 .10 80 60 40 20 0 50
E6: 23 .05 100 80 60 40 20 0
EOL 51 34.5 25 36.5 76 122.5

Pick the strategy corresponding to minimum expected regret.

i.e., buy 20 copies
 Perfect Information
Perfect Information would tell us with certainty which
outcome is going to occur

Having perfect information before making a decision

would allow choosing the best payoff for the outcome
EPPI (Expected payoff of perfect information)
EPPI = (probability of outcome)x ( best payoff of outcome)

Here, corresponds to the perfect situation when quantity

ordered is same as the demand.

EPPI= .05x 360 + .10x380+.30x400+.40x420+.10x440+

.05x460 = 411
 Expected Value of
Perfect Information (EVPI)
The amount by which perfect information would
increase our expected payoff
Provides an upper bound on what to pay for
additional information.
The worth of obtaining the perfect information
EVPI = EPPI EMV= 411-386=25=EOL

EPPI = Expected payoff with perfect information

EMV = Expected payoff without perfect information
 Caselet
Technico Ltd. has installed a machine costing Rs 4 lacs and is
in the process of deciding on an appropriate number of a
certain spare parts required for repairs. The spare parts cost
Rs 4,000 each but are available only if they are ordered now.
In case the machine fails and no spares are available, the cost
to the company of mending the plant would be Rs 18,000.
The plant has an estimated life of 8 years and the probability
distribution of failures during this time, based on experience
with similar machines, is as follows:
Fail. during 8 yrs 0 1 2 3 4 5 6+
Probability .1 .2 .3 .2 .1 .1 0
 Cost C = 4,000 S, when F<=S
= 4000 S+18000 (F-S) when F>S
Cost Matrix (in 000)
No. of Spares, Aj

Failures Prob A1:0 A2:1 A3:2 A4: 3 A5: 4 A6:5

E1: 0 .1 0 4 8 12 16 20
E2: 1 .2 18 4 8 12 16 20
E3: 2 .3 36 22 8 12 16 20
E4: 3 .2 54 40 26 12 16 20
E5: 4 .1 72 58 44 30 16 20
E6: 5 .1 90 76 62 48 34 20

Expected 41.4 29.2 20.6 17.4 17.8 20

Cost (EC)
Regret Matrix
No. of Spares, Aj

Failures Prob A1:0 A2:1 A3:2 A4: 3 A5: 4 A6:5

E1: 0 .1 0 4 8 12 16 20
E2: 1 .2 14 0 4 8 12 16
E3: 2 .3 28 14 0 4 8 12
E4: 3 .2 42 28 14 0 4 8
E5: 4 .1 56 42 28 14 0 4
E6: 5 .1 70 56 42 28 14 0

Expected 32.2 20 11.4 8.2 8.6 10.8

Regret (ER)
Note: regret matrix derived from cost matrix by taking the
difference of the least value in each row from other values of the
row.
ECPI= 9.2
EVPI=EC-ECPI=17.4-9.2=8.2 thousand rupees.
 Practice Problems
1) A physician purchases a particular vaccine on Monday of
each week. The vaccine must be used within the week
following, otherwise it becomes worthless. The vaccine
costs Rs 20 per dose and the Physician charges Rs 60 per
dose. In the past 50 weeks, the physician has
administered the vaccine in the following quantities:

Doses per week: 20 25 40 60

No. of weeks : 5 15 25 5
a) Prepare a payoff matrix
b) Obtain a regret matrix
c) Determine optimum number of doses the physician
should buy
d) Determine the maximum amount the physician would be
willing to pay per week for a perfect information about
the number of doses expected to be demanded in a week
2) A television dealer finds that the cost of a TV in stock for a
week is Rs 30 and the cost of unit shortage is Rs 70. For one
particular model of television the probability distribution of
weekly sales is as follows:

Weekly Sales: 0 1 2 3 4 5 6
Probability: .10 .10 .20 .25 .15 .15 .05

How many units per week should the dealer order?

Also, find EVPI.