Decision Analysis Notes

8.
0 DECISION ANALYSIS
8.1 Introduction
Decision analysis concerns with evaluating and choosing one or more best alternatives among
several existing alternatives. This analysis considers all the possible alternatives and possible
outcomes.
We first define the following concepts
 Decision alternatives: these are possible strategy or strategies that a decision maker
decide to take after the analysis.
 States of nature: the uncontrollable future events that can affect the outcome of a
decision.
 Payoff: the outcome measure such as profit, cost, etc. Each combination of a decision
alternative and a state of nature has an associated payoff.
 Payoff table: A tabular representation of the payoffs for a decision in a decision problem.
8.2 Steps in Decision Analysis

In order to analyze a problem and making an appropriate decision, the following steps are
followed
1. Clearly define the problem
2. List all possible alternatives
3. Identify all possible outcomes/states of nature for each alternative
4. Identify the payoff for each alternative & outcome combination
5. Use a decision modeling technique to choose an optimal alternative
To illustrate this, consider the Thompson Lumber problem

1. Decision: Whether or not to make and sell storage sheds
2. Alternatives:
• Build a large plant
• Build a small plant
• Do nothing
3. Outcomes: Demand for sheds will be high, moderate, or low
4. Payoffs: The following payoff table was obtained below
1
Outcomes (Demand)
Alternatives High Moderate Low
Large plant 200,000 100,000 -120,000
Small plant 90,000 50,000 -20,000
No plant 0 0 0
5. Apply one of the decisions modelling method.
8.3 Types of Decision Modeling Environments

There are three types of decision models. These are;
Type 1: Decision making under certainty
Type 2: Decision making under uncertainty
Type 3: Decision making under risk
8.3.1 Decision making under certainty
• The consequence of every alternative is known
• Usually there is only one outcome for each alternative
• This seldom occurs in reality
8.3.2 Decision making under uncertainty
In this model, probabilities of the possible outcomes are not known and hence not
involved in the analysis. Decision making methods include
• Maximax
• Maximin
• Equally likely
• Minimax regret
1. Maximax Criterion
• It is also known as optimistic approach
• Assume the best payoff will occur for each alternative
The Thompson’s lumber analysis using maximax criterion is shown below
2
Outcomes (Demand)
High Moderate Low Maximum

Alternatives Payoff
Large plant 200,000 100,000 -120,000 200,000
Small plant 90,000 50,000 -20,000 90,000
No plant 0 0 0 0
The maximum of the maxima payoffs is $200,000

Decision: Choose the large Plant (Best payoff)
2. Maximin criterion
 It is also known as pessimistic approach
 Assume the worst payoff will occur for each alternative
The Thompson’s lumber analysis using maximin criterion is shown below
Outcomes (Demand)
High Moderate Low Minimum

Alternatives payoff
Large plant 200,000 100,000 -120,000 -200,000
Small plant 90,000 50,000 -20,000 -20,000
No plant 0 0 0 0
Maximum of the minima payoffs is $0

Decision: Choose the No Plant (Best payoff)
3. Equally likely criterion

 Assumes all outcomes are equally likely and uses the average payoff
 Sum all payoffs of an alternative and divide by the number of outcomes
 The alternative with maximum average is preferable
The Thompson’s lumber analysis under equally likely criterion is shown below
3
Outcomes (Demand)
Alternatives High Moderate Low Average payoff
Large plant 200,000 100,000 -120,000 60,000
Small plant 90,000 50,000 -20,000 40,000
No plant 0 0 0 0
The maximum average is $60,000.

Decision: Choose the Large Plant (Best payoff)
4. Minimax Regret Criterion
 Regret or opportunity loss measures much better we could have done
 To compute the regrets, we consider the states of nature rather than alternatives.
Regret of an outcome = (best payoff) – (actual payoff)
The strategy is to minimize the amount of regret we would experience.
The Thompson’s lumber analysis under Minimax regret criterion is shown below
The following table gives regret values as well maximum regrets of each alternative
Outcomes (Demand)
Alternatives High Moderate Low Max. regret
Large plant 0 0 120,000 120,000
Small plant 110,000 50,000 20,000 110,000
No plant 200,000 100,000 0 200,000
Minimum of the maximum regret is $110,000

Decision: Choose small plant.
8.3.3 Decision Making Under Risk

This technique is used when the probabilities of outcomes are available
Two approaches are normally used
 Expected Monetary Value (EMV)
 Expected Opportunity Loss (EOL)
1. Expected Monetary Value (EMV)

 This technique uses the probabilities to calculate the average payoff for each
alternative. The following formula is used
4
EMV (for alternative i) = ∑(probability of outcome) x (payoff of an outcome)
 An alternative with the highest EMV is chosen.
The Thompson’s lumber analysis under Expected Monetary Value (EMV) is

shown below
EMV ( Large Plant )=200000 ×0.3+ 100000× 0.5+ (−120000 ) ×0.2
¿ 60,000+50,000−24,000=86,000
EMV ( Small Plant ) =90,000 ×0.3+50,000 ×0.5+ (−20,000 ) × 0.2
¿ 27,000+25,000−4,000=48,000
EMV ( No Plant )=0 × 0.3+0 ×0.5+ (−0 ) × 0.2=0
The following table summarizes
Outcomes (Demand)
Alternatives High Moderate Low EMV
Large plant 200,000 100,000 -120,000 86,000
Small plant 90,000 50,000 -20,000 48,000
No plant 0 0 0 0
Probability 0.3 0.5 0.2
Maximum EMV = $86,000

Decision: Choose large plant
2. Expected Opportunity Loss (EOL)

 It determines how much regret do we expect based on the probabilities?
 We first need to find a regret/opportunity loss table.
 The following formula is then used
EOL (for alternative i) = ∑(probability of outcome) x (regret of outcome)
 An alternative with the least EOL is preferable.
We illustrate this using Thompson Lumber’s problem
The regret table with probabilities is shown below, while the EOL are computed in similar
fashion as the EMV.
Outcomes (Demand)
Alternatives High Moderate Low EMV
5
Large plant 0 0 120,000 24,000
Small plant 110,000 50,000 20,000 62,000
No plant 200,000 100,000 0 110,000
Minimum EOL is $24,000

Decision: Choose large plant.
8.4 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.
8.4.1 Expected Value With Perfect Information (EVwPI)
This is the expected payoff of having perfect information before making a decision. This
measure is computed using the formula
EVwPI = ∑ (probability of outcome) x ( best payoff of outcome)
8.4.2 Expected Value of Perfect Information (EVPI)

The amount by which perfect information would increase our expected payoff. It
provides an upper bound on what to pay for additional information. It is computed using
the formula
EVPI = EVwPI – EMV
To illustrate this idea let us consider the Thompson’s Lumber problem
The bolded payoffs would be chosen based on perfect information (knowing demand
level)
Outcomes (Demand)
Alternatives High Moderate Low
Large plant 200,000 100,000 -120,000
6
Small plant 90,000 50,000 -20,000
No plant 0 0 0
From the payoff table we have,

EVwPI =200,000 ×0.3+100,000 × 0.5+0 ×0.2=$ 110,000
Since the EMV is known to be $86,000
Then,
EVPI=$ 110,000−$ 86,000=$ 24,000
Comment: The “perfect information” increases the expected value by $24,000
8.5 Decision Trees
It can be used instead of a table to show alternatives, outcomes, and payoffs. It consists of
nodes and arcs. It also shows the order of decisions and outcomes.
The decision nodes are represented by squares or rectangles and the outcome nodes by
circles or ovals. A decision Tree for Thompson Lumber problem is shown below;
8.5.1 Folding Back a Decision Tree

For identifying the best decision in the tree, we normally use EMV.
 Work from right to left
 Calculate the expected payoff at each outcome node
 Choose the best alternative at each decision node (based on expected payoff)
7
Consider the Thompson Lumber problem for illustration
Maximum EMV is $86,000 at node one.

Decision: Choose large plant.

Decision Analysis Notes

Uploaded by

Copyright:

Available Formats

Decision Analysis Notes

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Decision Analysis Notes

Uploaded by

Copyright:

Available Formats

8.

8.2 Steps in Decision Analysis

To illustrate this, consider the Thompson Lumber problem

Alternatives High Moderate Low

Large plant 200,000 100,000 -120,000

Small plant 90,000 50,000 -20,000

5. Apply one of the decisions modelling method.

8.3 Types of Decision Modeling Environments

High Moderate Low Maximum

Large plant 200,000 100,000 -120,000 200,000

Small plant 90,000 50,000 -20,000 90,000

The maximum of the maxima payoffs is $200,000

High Moderate Low Minimum

Large plant 200,000 100,000 -120,000 -200,000

Small plant 90,000 50,000 -20,000 -20,000

Maximum of the minima payoffs is $0

3. Equally likely criterion

Alternatives High Moderate Low Average payoff

Large plant 200,000 100,000 -120,000 60,000

Small plant 90,000 50,000 -20,000 40,000

The maximum average is $60,000.

Alternatives High Moderate Low Max. regret

Large plant 0 0 120,000 120,000

Small plant 110,000 50,000 20,000 110,000

No plant 200,000 100,000 0 200,000

Minimum of the maximum regret is $110,000

8.3.3 Decision Making Under Risk

1. Expected Monetary Value (EMV)

The Thompson’s lumber analysis under Expected Monetary Value (EMV) is

Alternatives High Moderate Low EMV

Large plant 200,000 100,000 -120,000 86,000

Small plant 90,000 50,000 -20,000 48,000

Probability 0.3 0.5 0.2

Maximum EMV = $86,000

2. Expected Opportunity Loss (EOL)

Alternatives High Moderate Low EMV

Small plant 110,000 50,000 20,000 62,000

No plant 200,000 100,000 0 110,000

Probability 0.3 0.5 0.2

Minimum EOL is $24,000

8.4.2 Expected Value of Perfect Information (EVPI)

Alternatives High Moderate Low

Large plant 200,000 100,000 -120,000

Probability 0.3 0.5 0.2

From the payoff table we have,

8.5.1 Folding Back a Decision Tree

Maximum EMV is $86,000 at node one.

You might also like