Please enable JavaScript.
Coggle requires JavaScript to display documents.
Decision Models, Bernardo Martín-Sánchez Hernández 00512844 - Coggle…
Decision Models
Performance Table Tool
Facilitates the decision making process
Organizes decision alternatives
Includes posible decision alternatives
States of nature
Events that may occur after making a decision
The intersections show the performance/yield
Performance represents gains or losses
Low Risk
More than one state of nature
Probability of each state can be estimated
Probabilities come from different things
Historical frequencies and data
Subjective estimates
Used when data isn't available
Calculations for the expected performance
Expected Yield = Σ (Yield × Probability)
The decision with the highest expected yield is always chosen
Decision Process
Series consisting of 3 steps
Step 2
A state of nature occurs
Once it occurs, the decision can't be changed
It isn't under our control
Step 3
The decision, combined with the state of nature generates a performance or a yield
Step 1
Select a decision alternative
It is under our control
Highest performance is chosen for gains
Lowest value is chosen for losses and costs
The decision depends on the ability to estimate the state of nature
Non Risk / Certainty
In this case, the state of nature is known
Probability of occurrence is well known
It is equivalent to having one state of nature
The decision with the best performance outcome is always selected
High Risk / Uncertainty
More than one state of nature
Probabilities are either unknown or can't be estimated
4 decision criteria
Maximin
Known as the pessimistic criterion
Finds the minimum performance for each decision
Selects the maximum of those minimums
Maximamax
Optimistic criterion
Finds the maximum performance for each decision
Selects the maximum of those minimums
Laplace
Assumes all states of nature are equally probable
Calculates expected performance and chooses the highest
Regret Minimax
Creates a regret table
Regret means the best performance for the state of nature, without considering its actual performance
Finds the maximum regret for each decision
Selects the decision with the smallest maximum regret
Regret = opportunity cost of not choosing the best decision
Bernardo Martín-Sánchez Hernández 00512844