DECISION ANALYSIS THEORY

Introduction

The Decision Process and Relevance of Operations Research

 Making appropriate decision is the most vital aspect in management.

 Everyone of us makes a number of decisions everyday. Some are important;

some are trivial; some decisions initiate a set of activities; some put an end to a
certain activities.
 In business environment, right decisions at the right times ensure success.

 This shows the importance of decision making.

 2

Introduction

 Problem is any variation between what was planned and what is actually

produced.
 Problem solving can be defined as:

“the process of identifying a difference between some actual and some desired
states of affairs and then taking action to resolve the difference.”
 3

Introduction

 Decision making requires for all human being because each of us makes

decision everyday in our life.

 Thus, decision making is universal.
 Decision making is a rational selection among alternatives.
 No alternative means, no decision required to be made.

 To put it in a nutshell, “decision making is the process of selecting or

choosing, based on some criteria, the best alternative among alternatives.”

 Decisions are easy. It’s only the rationale that is hard.
 4

The Decision Making Process

 Rational decision making goes through the following steps:

 Step 1Identify and define the problem/opportunity

 Problem is a necessary condition for making a decision, i.e. there would be

no need for decisions if problems did not exist.
 Step 2Define the set of alternative solutions

 Step 3 Determine the criteria to evaluate alternatives

 Identifying those characteristics that are important in making the decision
 Step 4Analyze the alternatives

 The advantages and disadvantages of each alternative, i.e. cost-benefit

analysis should be performed.
 5

The Decision Making Process

 Step 5 Select the best alternative
 Select the best alternative that suits to solve our decision problem.
 In selecting the best alternative, factors such as risk, timing and limiting
factors should be considered adequately.
 Step 6 Implement the solution
 Putting the decision into action
 Step 7 Establishing a control and evaluation system
 Ongoing actions need to be monitored; following the decision and
evaluate the results and determine if a satisfactory solution has been
obtained.
 6

 Just to recall, the previous chapters have focused mainly on decision making

when the consequences of alternative decisions are known with a reasonable

degree of certainty.
 This decision-making environment enabled formulating helpful mathematical models (linear
programming) with objective functions that specify the estimated consequences of any
combination of decisions.

 However, decisions often must be made in environments that are much more

fraught with uncertainty.

 7

Decision Analysis---

 Decision analysis provides a framework and methodology for rational

decision making when the outcomes are uncertain.

 In general terms, the decision maker must choose an action from a set of
possible actions.

 This choice of an action must be made in the face of uncertainty, because the
outcome will be affected by random factors that are outside the control of the
decision maker.
 8

Decision Analysis---

 These random factors determine what situation will be found at the time that
the action is executed.

 Each of these possible situations is referred to as a possible state of nature.

nature

 For each combination of an action and a state of nature, the decision maker
knows what the resulting payoff would be.

 The payoff is a quantitative measure of the value to the decision maker of the
consequences of the outcome.
 9

Decision Analysis---

 For example,
example the payoff frequently is represented by the net monetary gain
(profit), although other measures also can be used.

 If the consequences of the outcome do not become completely certain even

when the state of nature is given, then the payoff becomes an expected value
(in the statistical sense) of the measure of the consequences.

 A payoff table commonly is used to provide the payoff for each combination
of an action and a state of nature.
 10

Decision Analysis---
 To sum-up, decision analysis has four basic components. These are:

 Decision Maker- the sole individual assigned for decision making

 List of Decision Alternatives- are options available to the decision maker.
 State of Nature(s)- are a set of possible scenarios or circumstances under which a
decision is made. They are uncontrollable.
 Payoffs- are estimated values represented by positive or negative in the payoff table.
 The number of payoffs depends on:
 The number of decision alternatives and state of nature.
• I.e. number of payoffs = [# of decision alternatives] [# of state of nature].
 11

Decision Analysis---

 From this viewpoint, the decision analysis framework can be summarized as

follows:

a. The decision maker needs to choose one of the alternative actions.

b. Nature then would choose one of the possible states of nature.

c. Each combination of an action and state of nature would result in a

payoff, which is given as one of the entries in a payoff table.

d. This payoff table should be used to find an optimal action for the decision
maker according to an appropriate criterion.
 12

 Decision theory treats decisions against nature.

 This refers to a situation where the result (return) from a decision depends on

action of another player (nature).

 For example,
example if the decision is to carry an umbrella or not, the return (get wet
or not) depends on what action nature takes.

 Nature does not care what the outcome is.

 13

 The fundamental piece of data for decision theory problems is a payoff

table:
Decision State of Nature
Alternative 1 2 … m

d1 r11 r12 … r1m

d2 r21 r22 … r2m
… ... … … …
dn
rn1 rn2 … rnm

 The entries rij are the payoffs for each possible combination of decision and

state of nature.
 14

 The decision maker selects one of the possible decisions d1; d2; …; dn. Say di.

 After this decision is made, a state of nature occurs. Say state j.

 The return received by the decision maker is rij .

 The question faced by the decision maker is: which decision to select?

 The decision will depend on the decision maker's belief concerning what

nature will do, that is, which state of nature will occur.

 Different assumptions about nature's behavior lead to different procedures for

selecting "the best" decision.

 The Decision Making Environment
15
 Decisions are made in three basic conditions:

1. Decision making under Certainty

2. Decision making under Risk

3. Decision making under Uncertainty

 Certainty- A manager (decision-maker) knows for sure the state of

nature which will take place.
 Uncertainty- This is where the decision-maker knows only the possible
states of nature.
 Risk- This is where the decision-maker knows also the probability of
each state of nature.
 BASIC STEPS IN DECISION MAKING (Under Risk &Uncertainty)
Step 1 List the Feasible Decision Alternatives
16

Step 2 List the States of Nature.

 States of Nature:

 Are the most important (critical factors) future events that are not under the
control of decision-maker;
 But which are likely to affect the payoffs (benefits) that can be obtained
from the decision alternatives.

Step 3 Calculate the Payoffs of the Matrix.

Matrix
 Payoff Table:
Table
 Shows the amount of payoffs that can be obtained for each possible
decisions made (i.e. combination of state of nature and decision
alternatives.)
 17

1. Decision Making under Uncertainty

 At times a decision maker cannot assess the probability of occurrence

for the various states of nature.

 Uncertainty occurs when there exist several (i.e., more than one)
future states of nature but the probabilities of each of these states
occurring are not known.

 In such situations the decision maker can choose among several

possible approaches for making the decision. A different kind of
logic is used here, based on attitudes toward risk.
 18

 Different approaches to decision making under uncertainty include the

following:

1. The Maximax Criterion

2. The Maximin Criterion

3. The Minimax Regret Criterion

4. The Criterion of Realism /Hurwicz approach/

5. Equally Likely Approach

 19

 The optimistic decision maker may choose the alternative that offers the
highest possible outcome (the “maximax” solution);

 The pessimist decision maker may choose the alternative whose worst
outcome is “least bad” (the “maximin” solution);
solution

 The third decision maker may choose a position somewhere between

optimism and pessimism (“Hurwicz” approach);
approach

 Another decision maker may simply assume that all states of nature are
equally likely (the so-called “principle of insufficient reason”), set all values
equal to 1.0/n, and maximize expected value based on that assumption;
 20

 The fifth decision maker may choose the alternative that has the smallest
difference between the best and worst outcomes (the “minimax regret”
solution).

 Regret here is understood as proportional to the difference between

what we actually get, and the better position that we could have
received if a different course of action had been chosen.

 Regret is sometimes also called “opportunity loss.”

 The minimax regret rule captures the behavior of individuals who

spend their post decision time regretting their choices.
 Example for Decision Making under Uncertainty
21

 Problem:
Problem We have a Tape and CD manufacturing company. It is facing increased
demand beyond its capacity, what is the best (optimal) way of accommodating this
increasing demand or what is the optimal way of increasing supply to the market
based on the payoff matrix given in the table below? Payoff Matrix
Table

States of Nature Decision Alternatives

Expand Build Subcontract

High Demand $ 500,000 $ 700,000 $ 300,000

Moderate Demand $ 250,000 $ 300,000 $ 150,000
Low Demand -$ 250,000 -$ 400,000 -$ 10,000
Failure -$ 450,000 -$ 800,000 -$ 100,000

 N.B: This company measures payoffs as simple profits over the next five years.
 Example for Decision Making under Uncertainty
22

 Solution under the different criteria:

1. The Maximax Criterion..\Graphs\payoff table 1.docx

 This criterion is preferred by optimistic managers.

Step 1 Identify the maximum payoff that can be obtained for each decision
alternative. Expand Build Subcontract
Max-payoff $ 500,000 $ 700,000 $ 300,000

Step 2 The optimum choice is the alternative which yields the maximum
payoff in this list obtained in step 1.
 Decision: The optimum choice is to build a new plant. Because;
$700,000>$500,000>$300,000
 Example for Decision Making under Uncertainty
23

 Solution under the different criteria:

2. The Maximin Criterion-

 This criterion is preferred by pessimistic managers.

 Step1 Identify the minimum payoff that can be obtained for each decision

alternative. Expand Build Subcontract

Min-payoff -$ 450,000 -$ 800,000 -$ 100,000

 Step2 The optimum choice is the alternative which yields the

maximum payoff in this list obtained in step 1.
 Decision: The optimum choice is to Subcontract. Because; -$100,000>-

$450,000>-$800,000.
 Example for Decision Making under Uncertainty
24

 Solution under the different criteria:

3) The Minimax Regret Criterion

 Step 1 Obtain the regret table:

Regret Value= Max. Payoff of a state of nature – The payoff

 Step 2 Identify the max. regret value for each decision
alternative.
 Step3 The optimal choice is the one which yields minimum in the list,
which has been obtained in step 2.
 Example for Decision Making under Uncertainty
25

 For Example (cont.)

Step 1 Re-write the Decisions Alternative Table to come up with the Regret Table:
 In the 1st Row/1st SN/:

Regret Value for $500,000= 700,000 – 500,000 = $200,000

Regret Value for $700,000 = 700,000 – 700,000 = 0

Regret Value for $300,000 = 700,000 – 300,000 = $400,000

 In the 2nd Row/2nd SN/:

Regret Value for $250,000 = 300,000 – 250,000 = $50,000

Regret Value for $300,000 = 300,000 – 300,000 = 0

Regret Value for $150,000 = 300,000 – 150,000 = $150,000
 N.B: Follow the same procedure to obtain the regret values for the 3rd and 4th rows.
 Example for Decision Making under Uncertainty
26

Example (cont. Regret Table

States of Nature Regret Table
Decision Alternatives
Expand Build Subcontract

High Demand $ 200,000 0 $ 400,000

Moderate Demand $ 50,000 0 $ 150,000
Low Demand $ 240,000 $ 390,000 0
Failure $ 350,000 $ 700,000 0

Step 2 Write the maximum decision value for each alternative:

Expand Build Subcontract

$ 350,000 $ 700,000 $ 400,000

 Decision: The optimum choice is to Expand. Because;

$700,000>$400,000>$350,000.
 Example for Decision Making under Uncertainty
27

 Solution under the different criteria:

4. The Criterion of Realism/ “Hurwicz” approach/

 Step 1 Choose a value of α such that 0<α<1, α is an index of your

optimism about future.
 Relatively large value of α represents higher degree of optimism about the
most optimistic state of nature.
 Step 2 Calculate the measure of realism value for each decision
alternative as:

Measure of Realism= (α*max payoff) + ((1-α)*min payoff)

 Step3 The optimal choice is the decision alternative which yields maximum
measure of realism value.
 Example for Decision Making under Uncertainty
28

 Solution under the different criteria:

4. The Criterion of Realism/ “Hurwicz” approach/(Cont.)

Step 1 Take α = 0.7

Step 2 MoR Expand = (0.7*500,000) + (0.3*(-450,000))= 215,000

MoR Build= (0.7*700,000) + (0.3*(-800,000))= 250,000

MoR Sub-contract = (0.7*300,000) + (0.3*(-100,000))= 180,000

Step 3 Decision:
 The optimal choice is to Build a New Plant. Because;
250,000>215,000>180,000
 29

2. Decision Making under Risk

Nature of Risk-

 In decision making under risk one assumes that there exist a number of
possible future states of nature, Nj.

 Each Nj has a known (or assumed) probability Pj of occurring, and there

may not be one future state that results in the best outcome for all
alternatives Ai.
 30

 Different approaches to decision making under risk include the following:

1. Expected Value Criterion
2. The Criterion of Rationality
3. The Criterion of Maximum Likelihood
 31

1. Expected Value Criterion

 Step 1 List the feasible decision alternatives.
 When state of nature is given as a discrete random variable; feasible
decision alternatives are exactly the same as the states of nature.

Step 2 Obtain the probability of each state of nature using the past data.

Step 3 Using the cost and price data, obtain the conditional profit table.

Step 4 Using the probabilities, obtain the expected value of profits for
each decision alternative.

Step 5 Optimal choice is the decision alternative which yields maximum

expected profit.
 32

2) Criterion of Rationality
 Sometimes we may not trust the past data that are used to obtain
probabilities for states of nature.
 Under those circumstances we can use this criterion which assumes that all
states of nature are equally likely.
 We use this criterion when we have limited or unreliable data to calculate
probabilities.
 The rest of the steps are exactly the same as those of the Expected Value
Criterion.
 33

3. The Criterion of Maximum Likelihood

 We can use this criterion particularly when one of the states of nature has
significantly higher probability than others.
 In this case:

Step 1 We assure that this state of nature with the highest probability
will actually occur.

Step 2 Given this assumption we chose the decision alternative which

yields maximum conditional profits.
 34

 Example

 The Mid Town Food store stocks mangoes during early summer season. These

are flown-in from Meritt Island, Florida, each Monday and just be sold within
the week following.
 In the past, the store has been experiencing the following sales of mangoes:

Quantities Buyers Bought # of weeks this occurred

20 10

25 30

40 50

60 10

 Food Store buys mangoes for $2 and sells them for $4.
 35

Example (cont.)

Required:
 Given this, apply each of the following criteria to determine the optimal

quantity of mangoes that must be stocked per week:

A. Expected Value Criterion

B. Criterion of Maximum Likelihood

C. Criterion of Rationality
 36

Solution

A. Using Expected Value Criterion

Step 1 Obtain the probabilities of states of nature

Probabilities of Weekly Demand Quantities Buyers Bought
10/100 = 0.1 20

30/100 = 0.3 25
40
50/100 = 0.5
60
10/100= 0.1
 37
Solution (Cont.)

Step 2 Obtain the Feasible Stocking Actions (Decision Alternative)

N.B: Obtain exactly the number of quantity demanded from the supplier.

20 25 40 60

Step 3 Obtain the Conditional Profit table

 Obtain the weekly profit values for each possible combination of Stocks and

Demand:
Profit = TR – TC
TR = P * Q sold
TC = C * Q stocked
P = 4, and C = 2
 38
Solution (Cont.)
 Weekly profits when Stocks = 20units/week and Demand = 20;

Profit = TR – TC = (20*4) – (20*2) = 40

 Weekly profits when Stocks = 25units/week and Demand = 20;

Profit = TR – TC = (20*4) – (25*2) = 30

 Weekly profits when Stocks = 40units/week and Demand = 20;

Profit = TR – TC = (20*4) – (40*2) = 0

 Weekly profits when Stocks = 60units/week and Demand = 20;

Profit = TR – TC = (20*4) – (60*2) = -40

 N.B: Follow the same procedure to obtain the conditional profits for other

Stockings and Demands.

 39

Solution (Cont.)

Conditional Profit Table

Demand/States of Feasible Stocking Action /Decision Alternatives/
Probability Nature/ 20 25 40 60

0.1 20 $ 40 $ 30 0 -$ 40
0.3 25 $ 40 $ 50 $ 20 -$ 20
0.5 40 $ 40 $ 50 $ 80 $ 40
0.1 60 $ 40 $ 50 $ 80 $ 120
 40

Solution (Cont.)

Step 4 Calculate the Expected Values for each decision alternative:

..\Graphs\Conditional Profit Table 1.docx

E(п)20 = (0.1*40) + (0.3*40) + (0.5*40) + (0.1*40) = $40

E(п)25 = (0.1*30) + (0.3*50) + (0.5*50) + (0.1*50) = $48

E(п)40 = (0.1*0) + (0.3*20) + (0.5*80) + (0.1*80) = $54

E(п)60 = (0.1*-40) + (0.3*-20) + (0.5*40) + (0.1*120) = $22

 Decision

 The optimum choice is to stock 40 units because it yields max. expected

profit of $54.
 41

B. Using the Criterion of Rationality

 Taking ¼ = 0.25 as our probability
 Solution
Demand/States Feasible Stocking Action/Decision
of Nature/ Alternatives/
Probability 20 25 40 60

0.25 20 $ 40 $ 30 0 -$ 40
0.25 25 $ 40 $ 50 $ 20 -$ 20
0.25 40 $ 40 $ 50 $ 80 $ 40
0.25 60 $ 40 $ 50 $ 80 $ 120
 42

Solution (Cont.)
 Expected value of profits:

E(п)20 = (0.25*40) + (0.25*40) + (0.25*40) + (0.25*40) = $40

E(п)25 = (0.25*30) + (0.25*50) + (0.25*50) + (0.25*50) = $45

E(п)40 = (0.25*0) + (0.25*20) + (0.25*80) + (0.25*80) = $45

E(п)60 = (0.25*-40) + (0.25*-20) + (0.25*40) + (0.25*120) = $25

 Decision

 The optimal choice is to either stock 25 or 40 units because they yield the
maximum expected profit of $45.
 43

C. Using the Maximum Likelihood Criterion

 We take the highest probability of 0.5 from the conditional probability table
 If so, the Feasible Stocking Actions will be???
 44

Exercise
 Suppose that you are the sales manager of Migros supermarket chain in

Istanbul. You are trying to make a decision about the optimal number of
bottles of beer that should be stocked for each month. The cost of each bottle
to Migros is $6o and the beer is sold for a revenue of $95 per beer. For each
bottle that is unsold by the end of the week and returned back, Migros receives
$20.The sales data for the last 50 weeks is given as follows:
 45

Quantities Buyers Bought # of weeks this occurred

20 20
21 15
22 10
23 5

Required

A. Using the Expected Value Criterion, obtain the optimal number of bottles the

sales manager of Migros should stock per month.

B. Using the maximum Likelihood Criterion, obtain the optimal number of

bottles the sales manager of Migros should stock per month.