isom
Decision Tree Analysis
Overview: A decision tree allows decision makers to analyze choices and their potential outcomes sequentially, helpful in making decisions under risk.
Steps of Decision Tree Analysis
Define the Problem: Clearly understand the issue at hand.
Structure the Decision Tree: Create a visual representation based on a decision table.
Assign Probabilities: Estimate probabilities for each state of nature.
Evaluate Alternatives: For each alternative, assign estimated payoffs based on their outcomes.
Calculate Expected Monetary Values (EMVs): Compute EMVs for each node based on the probabilities and payoffs.
Elements of a Decision Tree
Decision Nodes: Illustrated as squares (□), representing choices and alternatives.
Example: Tom's Lamber can either:
Do nothing
Implement a small plan
Implement a large plan
State of Nature Nodes: Represented by circles (○), indicating possible outcomes of each decision.
Example outcomes might include favorable/unfavorable market conditions.
Expected Monetary Value Calculation
EMV is calculated by multiplying the probability of each outcome by its corresponding payoff.
For example, if a large plan yields:
Favorable Market: $200
Unfavorable Market: -$180
Probability of either market state = 0.5, then:
EMV = (0.5 * 200) + (0.5 * -180) = $10
Sequential Decision Making
In complex scenarios, decisions may depend on previous choices, visualized effectively through a tree structure.
Example with two decisions:
Conduct a market survey (cost: $10,000).
Based on survey results, make decisions for small or large plans based on favorable/unfavorable outcomes.
The probabilities derived from this survey will not be perfect; such analyses are influenced by prior probabilities and Bayesian inference.
Cost Analysis in Decision Making
Each decision incorporates tangible costs (e.g., a survey cost), affecting the final net payoff.
Payoff calculations must reflect these costs for accurate value assessments.
Value of Sample Information
The Expected Value of Perfect Information (EVPI) is determined by comparing:
The expected value when using sample (or imperfect) information, considering the cost.
COin a survey scenario, the results inform decisions under uncertainty but always yield some measure of prediction, never certainty.
EVSI = EV with sample information + cost - EV without sample information.
Utility Theory
Traditional monetary calculations may not encompass the full value of decisions due to personal preferences or risk tolerances.
Utility: A measure that incorporates subjective preferences in decision-making processes.
Example: In a gamble:
Accept a guaranteed $2 million or gamble for up to $5 million with a chance of losing everything.
Rational choices depend on individual risk tolerance; for a risk-averse person, guaranteed amounts may provide higher utility than risking for larger, uncertain payouts.
Conclusion
Decision trees facilitate complex decision analysis, especially in sequential scenarios with risks. Using them helps clarify computations of probabilities, outcomes, and personal utility preferences, ultimately guiding effective strategic choices.