DECISION ANALYSIS

By the end of this topic students are expected to be able to:

➢ Have a clear understanding decision making
➢ I dentify the decision making environments
➢ Identify the decision making criteria under each decision
environemnt
➢ Use the decision making criterion to evaluate business problems
➢ Be able to use decision trees in business decision making
What do we mean by decision?
• Decision making involves choosing the best course of action
among several alternatives.
• Decision making therefore involves choosing from a number of
alternatives, the best course of action
• In-order for the manager to make right decision, at the right time,
the possible courses of action and the possible outcomes
should be known
Important terms
• Action: A decision without options is not a decision, it is an action
• Alternative courses of actions: A decision maker must have two
or more possible choices to evaluate prior to selecting one course
of action from among the alternative courses of action
• Events (States of nature): Events are occurrences outside the
control of the decision maker.
• Payoff. In business applications, this payoff is usually expressed in
terms of profits or costs
• Outcomes. is a conditional value of an event.
• Probability distribution: gives the list of the events/States of
nature with the probabilities we would expect to see their
associated outcomes
The Decision-Making Environment
• Decision making is about the future. Three decision making
environments are normally identified for the purpose of formal decision
making:
• Certainty Environment: In this environment, only one state of nature
exists. The future is therefore, absolutely predictable. There is accurate,
measurable, reliable information on which to base our decision
• Uncertainty Environment: In this case, more than one state of nature
exists, but the decision maker has no knowledge about the various
states of nature or he/she does not have enough information to assist
him/her to assign probabilities to the different states of nature.
• Risk Environment. In this one, more than one state of nature exists, but
the decision maker, has information concerning the various states of
nature at hand that will assist in assigning probabilities to each of the
possible state of nature
Decision Making Under Uncertainty
• Four different decision criteria reflecting different attitudes of the
decision maker are herewith discussed including:

a) Pessimistic decision criteria

b) Optimistic decision criteria
c) Equally likely decision criteria
d) Opportunity loss decision criteria
Pessimistic Decision Criterion
• It assumes the worst is going to happen such that the strategy is to
choose the best among the worst.
• choose the maximum among the minimum profits, or the
minimum among the highest costs. The general procedure is:
a) For each possible strategy, identify the worst payoff value

b) Select the strategy with the best anticipated payoff value

Example:
Suppose we have the following pay-off matrix from four strategies and
four events (000 TZS) for 100, 200, 300 and 400 demand and supply
levels:
Event (Demand) E1 E2 E3 E4
Strategy (Supply) 100 200 300 400
S1 100 200 200 200 200
S2 200 100 400 400 400
S3 400 0 600 600 600
S4 400 -100 400 500 800
Pessimistic Decision Criterion

Events (Demand) E1 E2 E3 E4
100 200 300 400
Strategy (Supply) Decision
S1 100 100 100 100 100 100* (Optimal)
S2 200 50 200 200 200 50
S3 300 0 150 300 300 0
S4 400 -50 100 250 400 -50

Therefore, choose strategy S1, which signifies that minimum loss is incurred
 Optimistic Decision Criteria

Event E1 E2 E3 E4
Strategy 100 200 300 400 Best pay-offs
S1 100 100 100 100 100 100
S2 200 50 200 200 200 200
S3 300 0 150 300 300 300
S4 400 -50 100 250 400 400* (Optimal)

Therefore, choose strategy S4, which signifies that, maximum pay-off

Equally Likely Decision Criteria
• This is applied when we do not have enough reasons to decide
otherwise i.e. the decision maker has no sufficient knowledge
as to which event will occur.
• The optimal decision is to choose the best possible average of
expected payoff
• The procedure is:
a) Find the average or expected payoff by adding all the possible
payoffs
b) Divide the total payoff by the number of possible events
c) Select the strategy with the optimal value
Equally Likely Decision Criteria
Event E1 E2 E3 E4
100 200 300 400
Strategy Best pay-offs
S1 100 100 100 100 100 100
S2 200 50 200 200 200 152.5
S3 300 0 150 300 300 187.5 (Optimal)
S4 400 -50 100 250 400 175
Savage Opportunity Loss Decision Criterion
• The opportunity loss decision criterion, sometimes called the
savage minimax regret decision criterion, assures the decision
maker that the opportunities for payoff that he/she has missed (or
lost) because of the random occurrence of a possible
unfavourable event, will be as small value as possible.
Procedure:
a) Identify the best payoff value, under each event
b) For each event, subtract the actual conditional payoff value
from this best value. These results are the regret or opportunity
loss values for this event
c) Repeat steps (a) and (b) for all possible events
d) For each possible strategy, identify the worst/maxim regret
value. Record this number as a new column
e) Select the strategy with the smallest (minimum) anticipated
opportunity loss value
Decision Making under Risk Environment
Within the risk decision making environment there are many different decision situations. What differentiates
this environment to the uncertainty environment is the presence of some information that can help the decision
maker assign probabilities to all states of nature.
The approaches available to assist in decision making under the risk environment are as follows:
• Expected monetary value (EMV) criterion
• Expected Opportunity Loss (EOL) criterion
• Maximum Likelihood Criterion
• Rationale Criterion
 Example
• NMC Co ltd is engaged in selling a certain perishable product. Each
item costs TZS 1,500/- and is sold for TZS 2,500/-. Any item not sold
by the end of the dayis disposed off for TZS 1,000/-. The company
forecats the followng daily demand levels for the item
• 20,000 30,000 40,000 50,000 60,000
• If the most likely demand is 40,000 items and the demand for
20,000 units is half as likely, the demand for 30,000 units is three
quarters as likely, the demand for 50,000 units is one quarter as
likely and the demand for 60,000 is half as likely. Required
a) Use the EMV decision criterion to choose the optimal supply
b) Use the EOL to choose the optimal level
c) Use the maximum likelihood criterion to choose the supply
Expected Monetary Value (EMV) Criterion
• Procedure:
a) Develop the payoff matrix
b) Compute the expected value for each decision alternative
c) Select the decision that gives the highest excpected value
among those computed in stage (b) above
The solution for NMC Co Ltd in the preceding slideis as follows:
Probability distribution of demand levels
Demand level (x) Probability P(x)
20000 1/6
30000 1/4
40000 1/3
50000 1/12
60000 1/6
Total 1.0
Payoff matrix
Decision State of nature (Demand levels)
(Supply level) 20000 30000 40000 50000 60000
Probability [p(x)] 1/6 1/4 1/3 1/12 1/6
20000 20,000,000 20,000,000 20,000,000 20,000,000 20,000,000
30000 15,000,000 30,000,000 30,000,000 30,000,000 30,000,000
40000 10,000,000 25,000,000 40,000,000 40,000,000 40,000,000
50000 5,000,000 20,000,000 35,000,000 50,000,000 50,000,000
60000 0 15,000,000 30,000,000 45,000,000 60,000,000
 Expected Monetary Values for each decision are:
Decision level (Supply) EMV in TZS

20000 20,000,000

30000 27,500,000

40000 31,250,000

50000 30,000,000

60000 27,500,000

The optimal level to supply is 40,000 that gives the highest EMV
Expected Opportunity Loss
• The procedure for conducting a decision analysis using EOL is as
follows:
a) With the help of the payoff table, develop a regret matrix
(opportunity loss table)
b) Compute the expected opportunity loss for each decision
alternative
c) The optimal decision corresponds to that alternative with the
least expected opportunity loss
 Regret matrix for the problem
Decision States of Nature (Demand levels)
(Supply level) 20000 30000 40000 50000 60000
Probability 1/6 ¼ 1/3 1/12 1/6
20000 0 10,000,000 20,000,000 30,000,000 40,000,000
30000 5,000,000 0 10,000,000 20,000,000 30,000,000
40000 10,000,000 5,000,000 0 10,000,000 20,000,000
50000 15,000,000 10,000,000 5,000,000 0 10,000,000
60000 20,000,000 15,000,000 10,000,000 5,000,000 0
We obtain the following EOL values
Decision (Supply level) EOL in TZS
20000 18,333,333.3
30000 10,833,333,3
40000 7,083,333.3
50000 8,333,333.3
60000 10,833,333.3
The best decision using EOL is to supply 40,000 units with expected opportunity loss of TZS 7,083,333.3

The EMV and EOL decision criteria must lead to the same conclusion.
Maximum likelihood criterion
• Under this criterion, a strategy with high likelihood of happening is
the one taken. That being the case, for the NMC Co Ltd example,
the supply of 40,000 units has the highest chance of happening
and therefore will be chosen.
Rationale Approach
• This technique is applied when one wants to use methods like
EMV and EOL criteria to make a decision by virtue of their strength
but, at the same time, there is not enough information that can
help assign probabilities to evenets. The rationale approach
requires that every event be given an equal chance of occuring.
After that one can proceed with the steps as explained here.
 Rationale approach
Decision State of nature (Demand levels)
(Supply 20000 30000 40000 50000 60000
level)
Probability 1/5 1/5 1/5 1/5 1/5 Expected
[p(x)] Value
20000 20,000,000 20,000,000 20,000,000 20,000,000 20,000,000 20,000,000
30000 15,000,000 30,000,000 30,000,000 30,000,000 30,000,000 27,000,000
40000 10,000,000 25,000,000 40,000,000 40,000,000 40,000,000 31,000,000
50000 5,000,000 20,000,000 35,000,000 50,000,000 50,000,000 32,000,000
60000 0 15,000,000 30,000,000 45,000,000 60,000,000 30,000,000
Expected Value of Perfect Information (EVPI)
• If the management has the perfect information for a product, they
would supply exactly the demanded quantity as depicted in the
following table with bolded figures. When these figures are
subjected into probabilities they result to what is called, the
expected value under perfect information
Continued

Decision State of nature (Demand levels)

(Supply level) 20000 30000 40000 50000 60000

Probability 0.167 0.250 0.333 0.083 0.167

20000 20,000,000 20,000,000 20,000,000 20,000,000 20,000,000
30000 15,000,000 30,000,000 30,000,000 30,000,000 30,000,000
40000 10,000,000 25,000,000 40,000,000 40,000,000 40,000,000
50000 5,000,000 20,000,000 35,000,000 50,000,000 50,000,000
60000 0 15,000,000 30,000,000 45,000,000 60,000,000
continued
1 1 1
• 𝐸𝑉𝑃𝐼 = 20,000,000 + 30,000,000 + 40,000,000 +
6 4 3
1 1
50,000,000 + 60,000,000 = 𝑇𝑍𝑆 38,333,333.3
12 6
• Now EVPI = (Expected value under perfect Informatio) – (Expected
value under Imperfect Information
• Expected value under imperfect information is the maximum
expected monetary value under risk. (TZS 31,250,000) for the case
of NMC Co Ltd example
• Hence EPPI = 38,333,333.3 – 31,250,000 = TZS 7,083,333.3
Relationship between EMV and EO
EMV EOL EMV + EOL
20,000,000 18,333,333.3 38,333,333.3
27,500,000 10,833,333.3 38,333,333.3
31,250,000 7,083,333.3 38,333,333.3
30,000,000 8,333,333.3 38,333,333.3
27,500,000 10,833,333.3 38,333,333.3
Decision Making with decision tree
• A decision tree is a tool that portrays a sequency of decisions
followed by a sequency of events. It can be looked at as an
extended payoff table during times when a series of decisions are
to be made under a condition of risk.
• A decision tree is always drawn from left to right and is analyzed
from right to left.
• We use small squares in a decision tree when decisions take
place and we use small circles to represent situations when
events happen
Example
• A manufacturer is presented with a proposal for a new product and
must decide whether to develop it. The cost of the development is TZS
2m with a probability of 0.7 that the project will be successful. If the
product is successfully developed, the manufacturer must decide
whether to begin manufacturing the product at a high level or a low
level. If the demand is high, the incremental profits given a high level of
manufacturing is TZS 7m and given a low level of manufacturing is TZS
1.5m. If the demand is low, the incremental profit given high level of
manufacturing is TZS 1m, and given low level of e manufacturing is TZS
1.5m. These incremental profit values are before the subtraction of the
TZS 2m development cost, and are thus, are gross figures. The
probability that the market the market will be high is estimated at 0.40
and that it will be low is 0.60.
Required
• (a) Draw a decision tree for this product development problem
• (b) Advise the manufacturer on the decision to take