Statistical Decision Theory
Decision Theory represents a general approach to
decision making which is suitable for a wide range of
management decisions, including:
Capacity product and
planning service design
location equipment
planning selection
Product –mix Credit policies
Problem Formulation
• A decision problem is characterized by decision
alternatives, states of nature, and
resulting payoffs.
• The decision alternatives are the different
possible strategies the decision maker
can employ.
• The states of nature refer to future events, not
under the control of the decision maker,
which will ultimately affect decision results.
• States of nature should be defined so that
they are mutually exclusive and contain all
possible future events that could affect the
results of all potential decisions.
4
Payoff Tables
• The consequence resulting from a specific
combination of a decision alternative and a
state of nature is a payoff.
• A table showing payoffs for all combinations
of decision alternatives and states of nature is
a payoff table.
• Payoffs can be expressed in terms of profit,
cost, time, distance or any other appropriate
measure.
5
Fundamentals of decision theory
Decision States of Payoff
alternatives nature
Courses of An occurrence Quantitative
action or over which measure of
strategies decision maker the outcome
has no control
Three-egg omelette Problem
• You are preparing a three-egg omelette.
Having already broken two good eggs into the
pan, you are suddenly assailed by doubts
about the quality of the third. As yet
unbroken.egg, two things may happen: either
the egg is good or it is rotten.
Egg omelette problem
Acts (strategies)
Events Probability Break 3rd egg Break 3rd egg Throw away 3rd
into pan into saucer & egg
inspect
Third egg is good 0.9 3-egg omlette 3-egg 2-egg
omelette, one omelette, one
saucer to wash good egg
destroyed
Third egg is rotten 0.1 No egg omlette 2-egg 2-egg omelette
2 good eggs omelette, one
destroyed saucer to wash
Pay-offs
States of
nature
• Certainty - Environment in which
relevant parameters have known
values
• Risk - Environment in which
certain future events have
probable outcomes
• Uncertainty - Environment in
which it is impossible to assess
the likelihood of various future
events
Risk vs. Uncertainty
• Risk
– Must make a decision for which the outcome is not known with
certainty
– Can list all possible outcomes & assign probabilities to the
outcomes
• Uncertainty
– Cannot list all possible outcomes
– Cannot assign probabilities to the outcomes
• Certainty
-is an environment in which future outcomes or state of nature are
known.
• Eg: Investment in Bank FD, there is CERTAINTY
regarding FUTURE PAYMENTS on maturity
• Investment in shares is risky
• Investment in shares FETCHING returns higher than
FD in another 2 years, is uncertain
Criteria of decision making under
uncertainity
0ptimism(Maximax or Minimin)
Pessimism(Maximin or Minimax)
Equal probabilities(Laplace)
Coefficient of
optimism(Hurwicz)
Regret(Salvage)
Decision tree
• A decision tree is a network that exhibits
graphically the relationship between the different
parts of the complex decision process.
• It is a graphical model of each combination of
various acts and states of nature along with their
payoffs, and probability distribution.
• A decision tree analysis involves the construction
of a diagram that shows, at a glance, when
decisions are expected to be made in what
sequence, their possible outcomes, &
corresponding payoffs.
• A DT consists of nodes, branches, probability estimates and
pay-offs
• Three types of “nodes”
– Decision nodes - represented by squares (□) It
represents a point of action where a decision maker
must select one alternative course of action among the
available
– Chance nodes - represented by circles (Ο) It
indicates a point of time where the decision maker will
discover the response to his decision
– Terminal nodes - represented by triangles
(optional)
• Solving the tree involves pruning all but the best decisions
at decision nodes, and finding expected values of all
possible states of nature at chance nodes
• Create the tree from left to right
Decision table and tree
States of nature
Strategies State 1 State 2
Strategy 1 Outcome 1 Outcome2
Strategy 2 Outcome 3 Outcome 4
Outcome 1
1
outcome2
Outcome 3
2
outcome 4
Decision tree example
Stay comfortable and dry
Bear unnecessary trouble of carrying umbrella
Get wet and uncomfortable
Remain dry and comfortable
Flour mill : Payoff Table
THIS IS A PROFIT PAYOFF TABLE
Alternatives States of Nature
Low High
Small 8 8
Medium 5 15
Large -11 22
(Profits in LAKHS of Rs )
25
Expected Opportunity Loss (EOL)
• It is the opposite of EMV
• EOL is defined as the difference between the
highest profit or pay-off and the actual profit due
to choosing a particular course of action in a
particular state of nature
• The conditional opportunity loss (EOL) for a
particular course of action is determined by taking
the difference between the payoff value of the
most favourable course of action and some other
course of action.