Decision Analysis: EMV, Decision Trees, Bayes' Rule, and Value of Information

0.0(0)
Studied by 0 people
call kaiCall Kai
learnLearn
examPractice Test
spaced repetitionSpaced Repetition
heart puzzleMatch
flashcardsFlashcards
GameKnowt Play
Card Sorting

1/16

flashcard set

Earn XP

Description and Tags

Vocabulary flashcards covering key definitions, mathematical formulas, and principles from Albright DADM Chapter 6 on decision criteria, decision trees, Precision Tree, multistage problems, Bayes' rule, and the expected value of information.

17 Terms

1
New cards

Expected Monetary Value (EMV)

A weighted average of the possible payoffs for a decision, weighted by the probabilities of the outcomes.

2
New cards

EMV Criterion

A decision approach that assesses probabilities for each outcome of each decision and selects the decision with the largest EMV, a practice sometimes referred to as "playing the averages."

3
New cards

Sensitivity Analysis (Spreadsheet EMV)

A method easily performed in a spreadsheet to evaluate how changes in input parameters, such as probabilities and payoffs, affect the calculated EMVs and the optimal decision.

4
New cards

Decision Tree

A graphical tool developed to represent decision problems—particularly complex ones—by clearly displaying the sequence of events (decisions and outcomes), probabilities, and monetary values.

5
New cards

Precision Tree

A powerful Excel add-in developed by Palisade Corporation that enables users to draw and label decision trees, run sensitivity analysis on key parameters, and automatically execute the folding-back procedure.

6
New cards

Folding-Back Procedure

An analytical procedure performed automatically by Precision Tree to evaluate decision tree nodes and determine optimal decisions.

7
New cards

Multistage Decision Problems

Real-world decision problems that evolve through time across multiple sequential stages rather than occurring in a single step.

8
New cards

Contingency Plan

An EMV-maximizing strategy in multistage problems that designates the first-stage decision and establishes subsequent decisions to be made once uncertain outcomes become known.

9
New cards

Bayes' Rule

A formal mathematical mechanism for updating probabilities as new information becomes available, which can be implemented using frequencies (counts) or probabilities.

10
New cards

Prior Probabilities

The original set of outcome probabilities assessed before any new information is observed, denoted as P(O1)P(O_1) through P(On)P(O_n) and summing to 11.

11
New cards

Posterior Probabilities

The updated set of outcome probabilities, denoted as P(Oi∣I)P(O_i|I), obtained via Bayes' rule after observing new information II, which can subsequently serve as priors for later information.

12
New cards

Conditional Probability

The probability of an outcome OO given that new information II has been observed, formally designated as P(O∣I)P(O|I).

13
New cards

Likelihoods

In Bayesian terminology, the conditional probabilities of observing information outcome II given specific outcomes O1O_1 to OnO_n, expressed as P(I∣O1)P(I|O_1) to P(I∣On)P(I|O_n).

14
New cards

Law of Total Probability

The denominator in Bayes' rule representing the total probability of an information event II, calculated as P(I)=P(I∣O1)P(O1)+...+P(I∣On)P(On)P(I) = P(I|O_1)P(O_1) + \text{...} + P(I|O_n)P(O_n).

15
New cards

Bayes' Rule (Two-Outcome Formula)

The formula for Bayes' rule when there are only two outcomes (OO and Not O\text{Not } O): P(O∣I)=P(I∣O)P(O)P(I∣O)P(O)+P(I∣Not O)P(Not O)P(O|I) = \frac{P(I|O)P(O)}{P(I|O)P(O) + P(I|\text{Not } O)P(\text{Not } O)}.

16
New cards

Expected Value of Information (EVI)

The monetary amount a firm is willing to pay to reduce uncertainty about an outcome, computed as EVI=EMV with (free) information−EMV without information\text{EVI} = \text{EMV with (free) information} - \text{EMV without information}.

17
New cards

Expected Value of Perfect Information (EVPI)

The maximum amount any information could be worth regardless of its form or accuracy, computed as EVPI=EMV with (free) perfect information−EMV without information\text{EVPI} = \text{EMV with (free) perfect information} - \text{EMV without information}.