[go: up one dir, main page]
Scribd
Upload
130 views6 pages

NOTES Decision Making Under Uncertainty

The document discusses various approaches to decision making under uncertainty, including maximizing expected monetary value and minimizing expected regret. It outlines criteria for evaluating decisions like Maximax, Maximin, Laplace, Minimax, and Hurwitz. Examples are provided to illustrate how to apply these criteria to decisions involving uncertain outcomes like selecting a warehouse size based on uncertain future demand levels.

Uploaded by

Inés
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
130 views6 pages

NOTES Decision Making Under Uncertainty

The document discusses various approaches to decision making under uncertainty, including maximizing expected monetary value and minimizing expected regret. It outlines criteria for evaluating decisions like Maximax, Maximin, Laplace, Minimax, and Hurwitz. Examples are provided to illustrate how to apply these criteria to decisions involving uncertain outcomes like selecting a warehouse size based on uncertain future demand levels.

Uploaded by

Inés
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
You are on page 1/ 6

<OPERATIONS MANAGEMENT L4: DECISION MAKING UNDER UNCERTAINTY

Overview

Goals: To be able to make decisions when there is uncertainty in the outcome of those
decisions; and to be able to structure and analyse complex, probabilistic decisions.
- Decision making process
- Decision making under uncertainty
- Decision making under risk

Decision-making process

• Define the problem


• Identify the set of alternatives
• Determine the criteria for making a decision
• Evaluate the alternatives using the criteria
• Choose the alternative

Decision types

• Certain decisions: Values (such as costs, capacity, no. of employees) that


describe a situation are known.
• Uncertain decisions: Certain values have unknown or random outcomes and
we have no information of the distribution that drives the outcome. For
example, demand for a completely new product
• Risky decisions: Certain values have probabilistic outcomes, but the distribution
of the probability is known. For example, demand for a product with a long
history
• Competitive decisions: Decisions taken in an environment where the outcome
of the decision depends on the (simultaneous thus unknown) choices by
somebody else

Decision making under uncertainty

Before we start, we need to identify...


• What are the Decision Alternatives that a decision maker can take?
• What are the States of Nature? (i.e. future events which the decision maker
cannot control)
• What is the pay-off for each decision alternative and state of nature?
• Payoff or profit (the outcome that results from a specific decision
alternative and the occurrence of a particular state of nature)
– What do I value in my decision?
– To obtain the maximum profit. To avoid large losses. To avoid the regret
from making the wrong decision. To make a logical, mathematical
decision?
How do you make decisions?

• Let us play a game of lottery


• Suppose you can choose between two lotteries.
• Here are their payoffs

Decision State of nature


Choices Good luck Bad luck
Lottery A £2000 £1000
Lottery B £3000 -£2000
Probability 0.8 0.2

• Which one would you choose?

Decision rules

• Maximax. Determine the best possible payoff for each scenario and choose the
alternative with the highest maximum payoff. An optimistic approach, but high
losses that are risked in order to access the highest payoff are completely
ignored
• Maximin. Determine the worst possible payoff for each scenario and choose
the “best worse” alternative with the largest minimum payoff. A pessimistic
(conservative) approach that maximises the “guaranteed minimum return”.
Risk adverse to minimise damage
• Laplace. Determine the average payoff for each option and choose the
alternative with the best average benefit. Assume that each of the states of
nature are equally as probable. The logical/mathematical choice
• Minimax. Identify the possible 'regret' or 'opportunity loss' for each alternative
under each state of nature and choose the alternative with the smallest
maximum regret (the “best worst” regret). Sore losers try to minimise the lost
opportunity from making the wrong decision Regret = (The best payoff under
the given state of nature) – (The actual payoff for the given alternative under
the same state of nature)
• Hurwitz. Select a co-efficient of optimism, 0 £ a £ 1 . If you are very optimistic
about the future then select near 1, pessimistic, select near 0. The rank
each option with
Hurwicz payoff, H     maximum payoff    1      minimum payoff 
• Maximum Likelihood. Choose the option with the highest pay-off for the state
of nature that is deemed most likely to occur.

Example: BP´s investment decision

• British Petroleum (BP) has some funds to invest in developing a new oil field.
However, due to recent events it can only afford to make one investment. The
choices and pay-off’s, depending on the size of the oil find are described in the
pay-off table below.

Decision Quantity of oil extracted from field


Small Med Large
North Sea 20 25 30
Azerbaijan -10 25 45
Gulf of Mexico -20 40 70
Probability 0.2 0.3 0.5

• Identify which decision would be taken with each of the “decision making
under uncertainty” criteria: Maximax; Maximin; Laplace; Minimax; Hurwitz
(with = 0 . 7 ); and Maximum Likelihood.

Exercise: Selecting the most appropriate size warehouse

• A company has four options for renting a new warehouse;


– small, medium, large and extra-large.
• The annual rent for each warehouse is
– small warehouse = £50, medium warehouse = £100, large warehouse =
£150 and extra-large warehouse = £200.
• The income that could be generated from each warehouse depends on the
level of the future demand.
– The probabilities of level of future demand are not known, thus we are
making a decision under uncertainty.
• Company experts forecast the income for each of the high, med & low demand
levels for each warehouse as follows

Cost of Income for future demand level


Alternatives alternative
(Rent) High Medium Low

Small £50
£80 £75 £70
warehouse

Medium £100
£150 £139 £110
warehouse

Large £150
£250 £180 £110
warehouse

XL £200
£320 £210 £100
warehouse
• Which warehouse would you select under each of the six criteria (Maximin,
Maximax, Laplace, Minimax, Hurwitz, Maximum Likelihood) if you considered
the pay-off to be the income minus the warehouse cost?
– You choose the confidence co-efficient in the Hurwitz criterion
– Assume that the Medium demand is the most likely to occur

Warehouse selection: a decision under uncertainty

Hurwitz criterion

• Select a “co-efficient of optimism”, 0 < α < 1. If you are very optimistic about
the future then select α near 1, pessimistic, select α near 0. Then you rank each
option with
Hurwicz payoff, H     maximum payoff    1      minimum payoff 
• Say we select α = 0.6.
• Then we select a large warehouse.

Maximum Likelihood

• Choose the option with the highest pay-off for the state of nature that is
deemed most likely to occur (medium).
• Let’s say that we believe that the state of nature most likely to occur is that
demand will be of a medium level.
• We select the Medium warehouse as it has the highest payoff for a medium
demand.
Critique of decision making under uncertainty

• Five of the criteria are irrational as they all (except the Laplace) ignore most of
the available information in the payoff matrix
• Consider the following payoff matrices

Decision making under risk

• Probabilistic (decision making with probabilities)


– Sometimes the states of nature can be assigned probabilities that
represent their likelihood of occurrence
– For decision problems that occur more than once, we can often
estimate these probabilities from historical data
• Decision rules
– Expected Monetary Value (EMV),
– Expected Opportunity Loss (EOL),
– Expected Value of Perfect Information (EVPI)

Decision rules

• Expected monetary value (EMV): Make a decision that maximises the EMV
(applied to a payoff matrix)
EMVi   j 1 p j wij
n

w
= payoff for alternative i under the jth state of nature
ij

p
= the probability of the jth state of nature
j

• Expected regret or opportunity loss (EOL): Make a decision that minimises the
EOL (applied to a regret matrix)
n
E O Li =å j =1
p j rij
rij
= regret for alternative i under the jth state of nature
p j = the probability of the jth state of nature
• Expected value of perfect information (EVPI)
– Suppose we could hire a consultant who could predict the future with
100% accuracy
– With such perfect information we could make each decision that
maximise the payoff: Expected Profit of a Perfect Predictor (EPPP)
– EPPP = Sum of the product of the highest profit and probability of each
state of nature.
EVPI = EPPP– EMV
– EVPI is the maximum amount of money we are willing to pay to improve
our knowledge of the future

Example:

You might also like