MGMT20005 Business Decision Analysis - Lecture 3: Value of Information

Vocabulary flashcards covering key terms and concepts from Lecture 3, including perfect information, sample information, Bayes' theorem, EVPI, EVSI, prior/posterior probabilities, and risk profiles.

Expected Value of Information

The value determined by the impact that acquiring information has on future actions, measuring whether it is worthwhile to acquire information to reduce uncertainty and risk.

Perfect Information

Information from a source or expert that is always 100% correct, defined mathematically as $P(\text{Outcome}_i \mid \text{Expert Says 'Outcome}_i \text{ Occurs'}) = 1$.

Expected Value of Perfect Information (EVPI)

The maximum value of information pertaining to an uncertainty, calculated as $EVPI = |EVwPI - EVwoPI|$, serving as a benchmark for evaluating sample information.

EVwPI

The expected value with perfect information about the state of nature, calculated as $EVwPI = \sum_{j=1}^{N} P(s_j) \max_i V_{ij}$.

EVwoPI

The expected value without perfect information about the state of nature, calculated as $EVwoPI = \max_i EV(d_i) = \max_i \sum_{j=1}^{N} P(s_j) V_{ij}$.

Sample Information

Additional information about states of nature obtained through experiments, studies, or market research, used to revise or update prior probabilities.

Prior Probabilities

The preliminary probability assessments for states of nature that represent the best probability values available to a decision maker prior to acquiring sample information.

Posterior Probabilities

The revised probabilities for states of nature obtained by updating prior probabilities with new sample information using Bayes' theorem.

Expected Value of Sample Information (EVSI)

The expected payoff gain from acquiring sample information, calculated as $EVSI = |EVwSI - EVwoSI|$, used to evaluate whether gathering extra information before deciding is worthwhile.

EVwSI

The expected value achievable when utilizing sample information about the states of nature to guide decision choices.

EVwoSI

The expected value achievable without obtaining or using sample information about the states of nature.

Bayes' Theorem

A formula describing the probability of an event based on prior knowledge and new conditional evidence, expressed as $P(B \mid A) = \frac{P(A \mid B)P(B)}{P(A \mid B)P(B) + P(A \mid \bar{B})P(\bar{B})}$ or $P(B \mid A) = \frac{P(A \cap B)}{P(A)}$.

Joint Probability

The probability of two events occurring together, computed in decision analysis by multiplying a prior probability by a conditional probability, $P(A \cap B) = P(A \mid B)P(B)$.

Conditional Probability

The probability of observing a specific sample outcome given that a particular state of nature exists, such as $P(s_F \mid s_1)$.

Folding-Back Approach

A backward pass method through a decision tree that calculates expected values at chance nodes and selects optimal branches with the maximum expected values at decision nodes.

Risk Profile

A representation showing all possible payoff outcomes alongside their associated probabilities for a specific decision strategy.