📈Theoretical Statistics Unit 12 – Decision theory

Key Concepts and Foundations

• Decision theory provides a formal framework for making rational decisions under uncertainty
• Involves identifying the available options, their potential outcomes, and the associated probabilities and utilities
• Incorporates both objective information (probabilities) and subjective preferences (utilities) to guide decision-making
• Draws upon various disciplines, including probability theory, statistics, economics, and psychology
• Aims to optimize the expected value or utility of a decision, considering the decision-maker's goals and risk attitudes
  • Expected value is the weighted average of possible outcomes, where weights are the probabilities of each outcome occurring
  • Utility represents the subjective value or satisfaction derived from a particular outcome
• Distinguishes between decisions under risk (known probabilities) and decisions under uncertainty (unknown probabilities)
• Provides a systematic approach to dealing with complex, high-stakes decisions in various domains (business, healthcare, public policy)

Decision-Making Framework

• The decision-making process typically involves several key steps:
  1. Identifying the decision problem and objectives
  2. Generating a set of alternative courses of action
  3. Evaluating the consequences and probabilities associated with each alternative
  4. Selecting the alternative that best aligns with the decision-maker's objectives and preferences
• Objectives should be clearly defined, measurable, and aligned with the decision-maker's goals and values
• Alternatives should be mutually exclusive, collectively exhaustive, and comparable in terms of their potential outcomes
• Consequences can be evaluated in terms of monetary values, health outcomes, environmental impacts, or other relevant metrics
• Probabilities can be estimated using historical data, expert judgment, or mathematical models
• The selected alternative should maximize the expected value or utility, subject to any relevant constraints or risk tolerances
• Sensitivity analysis can be conducted to assess the robustness of the decision to changes in key assumptions or parameters

Probability Theory in Decision Analysis

• Probability theory provides a mathematical framework for quantifying and reasoning about uncertainty

• Probabilities are assigned to events or outcomes, representing the likelihood of their occurrence

• Joint probability refers to the probability of two or more events occurring simultaneously, denoted as $P(A \cap B)$

• Conditional probability is the probability of an event occurring given that another event has already occurred, denoted as $P(A|B)$

• Bayes' theorem relates conditional probabilities and can be used to update probabilities based on new evidence:

  $P(A|B) = \frac{P(B|A)P(A)}{P(B)}$

• Independence implies that the occurrence of one event does not affect the probability of another event

• Probability distributions (discrete or continuous) describe the likelihood of different outcomes for a random variable

• Expected value is a key concept in decision analysis, representing the average outcome weighted by probabilities

Utility Theory and Preferences

• Utility theory captures the subjective preferences and risk attitudes of the decision-maker
• Utility functions assign numerical values to outcomes, reflecting their relative desirability
• Common utility functions include:
  • Linear utility: $U(x) = ax + b$, where $a$ and $b$ are constants
  • Exponential utility: $U(x) = -e^{-ax}$, where $a$ is a risk aversion coefficient
  • Logarithmic utility: $U(x) = \log(x)$
• Risk attitudes can be classified as risk-averse, risk-neutral, or risk-seeking, based on the shape of the utility function
• Certainty equivalent is the guaranteed amount that a decision-maker would accept in lieu of a risky prospect
• Expected utility is the weighted average of the utilities of possible outcomes, where weights are the probabilities
• Maximizing expected utility is a common objective in decision analysis, accounting for both probabilities and preferences

Decision Trees and Expected Value

• Decision trees provide a graphical representation of the decision problem, depicting the sequence of decisions and chance events

• Nodes in a decision tree represent decision points (squares) or chance events (circles), while branches represent the available options or possible outcomes

• Probabilities and payoffs (or utilities) are assigned to each branch, based on the available information and preferences

• The expected value (EV) of a decision alternative is calculated by multiplying the probability and payoff of each possible outcome and summing the results:

  $EV = \sum_{i=1}^{n} p_i v_i$

  where $p_i$ is the probability and $v_i$ is the payoff of the $i$-th outcome

• Folding back the tree involves calculating the expected values at each chance node and selecting the alternative with the highest EV at each decision node

• Sensitivity analysis can be performed by varying the probabilities or payoffs to assess the robustness of the optimal decision

Bayesian Decision Theory

• Bayesian decision theory combines probability theory and utility theory to make decisions under uncertainty

• Prior probabilities represent the initial beliefs about the likelihood of different states of nature, before observing any data

• Likelihood functions specify the probability of observing a particular set of data given each possible state of nature

• Posterior probabilities are updated using Bayes' theorem, incorporating both the prior probabilities and the observed data:

  $P(\theta|x) = \frac{P(x|\theta)P(\theta)}{P(x)}$

  where $\theta$ is the state of nature and $x$ is the observed data

• Bayesian inference allows for the continuous updating of probabilities as new information becomes available

• Bayesian decision rules select the alternative that maximizes the expected utility, calculated using the posterior probabilities

• Conjugate priors are chosen to simplify the calculation of posterior probabilities, as they result in posterior distributions from the same family as the prior

Risk and Uncertainty in Decision-Making

• Risk refers to situations where the probabilities of different outcomes are known, while uncertainty refers to situations where the probabilities are unknown or not well-defined
• Risk aversion implies a preference for a certain outcome over a risky prospect with the same expected value
• Risk premium is the amount that a risk-averse decision-maker is willing to pay to avoid a risky situation
• Ambiguity aversion refers to the preference for known risks over unknown risks, even when the expected values are the same
• Robust decision-making approaches aim to identify alternatives that perform well across a wide range of possible scenarios, rather than optimizing for a single best-guess scenario
• Sensitivity analysis and scenario planning are tools used to assess the impact of uncertainty on decision outcomes
• Decision-making under uncertainty may involve the use of decision criteria such as maximin (pessimistic), maximax (optimistic), or minimax regret

Applications and Case Studies

• Medical decision-making: Diagnosis, treatment selection, and resource allocation based on patient characteristics, treatment effectiveness, and costs
• Business decisions: Investment appraisal, product development, pricing strategies, and market entry decisions based on market conditions, competitor actions, and consumer preferences
• Environmental policy: Choosing among alternative regulations or interventions to address environmental risks (climate change, air pollution) based on scientific evidence, economic impacts, and societal values
• Legal decision-making: Jury verdicts, plea bargaining, and sentencing decisions based on the strength of evidence, the severity of the offense, and the likelihood of recidivism
• Personal finance: Retirement planning, insurance purchasing, and portfolio allocation decisions based on financial goals, risk tolerance, and market expectations
• Transportation planning: Selecting among alternative infrastructure investments or traffic management strategies based on travel demand, construction costs, and environmental impacts
• These case studies demonstrate the wide-ranging applicability of decision theory to real-world problems involving uncertainty, trade-offs, and conflicting objectives