Theory for the Exam Flashcards
Difference between lexicographic approach and Additive method
Lexicographic approach is logical and simple but does not allow for trade offs the same way the additive method does.
The additive method also considers the DM risk preference
Five Os Method
Useful checklist for grouping and clustering Owner - Decision maker, Shareholders Objectives - criteria Options - alternatives Occasions - context, constraints Odds - uncertainties
Define Utility
The amount of satisfaction that a consumer receives by consuming a good or service
Helps quantify the DM risk preference
Von Neumann and Morgenstern Axioms
5 in total Complete Ordering Continuity Independence Unequal Probability Compound Lottery
Order Axiom (Complete Ordering)
If r1 > r2
and r2 > r3
then r1 > r3
Independence Axiom
Given r1 and r2 are indifferent and any reward r3 and probability p, the following lotteries are indifferent
Posterior Probabilities
flipping of the formulas you already have
Significance of zeros in Markov chain matrix
Represent transitions that cannot be made. It is impossible to go from one to the other
Simons 1976 Satisficing model
Simon 1976
Set a satisfactory level of performance for the most important attribute
Eliminate alternatives
Repeat steps for next most important alternative
Can model on an graph with criteria on each axis
Satisfice because unable to maximise
Useful at preliminary stages and routine decisions
Based on bounded rationality
is NOT a prescriptive model
Risk taking
Convexed
CME > EMV
Risk Averse
Concaved
CME < EMV
5 criteria for additive function
5 in total Preferential Independence Complete Compensation Corresponding trade off Interval scale, Scaling constants
Maximax Benefit
The “highest” maximum Benefit
Risk taking
Optimistic
Maximin
The “highest” minimum benefit
Risk Averse
Pessimistic
Minimax Regret Criteria
The “lowest” maximum regret
Opportunity loss
Pessimistic
Unichain
There is only one single recurrent class. All transient states transition into this.
Aperiodic
If t = 0 or no such a t can be found
Possible to return to itself with a positive probability
Completely ergodic
Unichain and Aperiodic
Local Scale
Defines possible outcomes in relation to the available alternatives given in the problem
Lazy approach
Global Scale
Using all possible outcomes in similar contexts, outcomes which could realistically occur
Hurwicz Criterion
Degree of optimism
Realistic view
Laplace Criterion
All states are equally likely
Continuity Axiom
If r1 > r2 and r2 > r3, then there exists p which makes a lottery of r1 (which will defo occur) indifferent to a lottery of r2 and r3
State is Accessible?
probability going from one to another is greater than 0
