Summary
Highlights
The video introduces basic quantitative decision analysis, emphasizing its simplicity and step-by-step approach for beginners or those seeking a review. It aims to explain 'what' and 'why' behind the concepts.
Decision analysis is illustrated with everyday scenarios, such as choosing a driving route to avoid traffic, and business-focused examples like investment allocation for an investment services firm, and evaluating mining operations for a mining company.
A mining company, Luster Incorporated, faces a decision: mine land with a 20% chance of profitable gold deposit (investment $80,000, profit $600,000) or sell the land for an immediate $75,000 profit, eliminating risk. This scenario forms the basis for the decision analysis.
The video explains the components of a decision table: alternative decisions (mine or sell), states of nature (gold or no gold), prior probabilities (20% for gold, 80% for no gold), and payoffs for each combination of decision and state of nature.
The Maximax Criterion, used by 'hopeless optimists,' aims for the best possible outcome without considering probabilities. In the mining example, this would be to mine the land, hoping for a $600,000 profit if gold is found.
The Maximin Criterion, for 'hardened pessimists,' focuses on choosing the best among the worst possible outcomes. In the mining example, this means selling the land for $75,000 to avoid the $80,000 loss if mining yields no gold.
This criterion considers the most likely state of nature based on given probabilities. Since 'no gold' is 80% likely, the decision would be to sell the land for $75,000, as it's the best outcome in that most probable state.
The Equally Likely Criterion assumes all states of nature have equal probability. For the mining example, with two states, each gets a 0.5 probability. Calculating the weighted average, mining the land yields $260,000, while selling yields $75,000, making mining the preferred choice under this criterion.
EMV uses the actual probabilities of each state of nature to calculate the expected payoff. For the mining example, mining the land has an EMV of $56,000, while selling the land has an EMV of $75,000, making selling the better decision based on the geologist's probabilities.
EVPI determines the maximum amount a company should pay for perfect information. It's calculated as the expected value under certainty minus the maximum EMV. For Luster Incorporated, the EVPI is $115,000, representing the theoretical maximum they'd pay for a perfect forecast.