Womack Report

Chapter 3 and 4.In any decision analysis, three factors need to be accounted for. There must be alternative courses of action, the state of nature needs to be listed for each alternative, and the payoff for each alternative must be listed.

Types of Decision-Making Environments:

1. Type 1: Decision-making under certainty — All relevant information is known.
2. Type 2: Decision-making under risk — There is unknown information, but probabilities can be estimated
3. Type 3: Decision-making under uncertainty — There is unknown information, and probabilities are unknown

Decision Strategies

1. Maximax — choose the alternative that maximizes the maximum payoff for every alternative. Maximax decisions are based on finding the greatest possible gain, and choosing that alternative. Optimistic decision making strategy.
2. Maximin — Choose the alternative with the maximum payoff in the worst case scenario. Maximin decisions are based on minimizing risk. Pessimistic decision making strategy.
3. Criterion of Realism (Hurwicz) — Use a weighted average based on optimism about results. CR = alpha * (row max) + (1- alpha) * (row minimum). This method only selects from the best and worst payoffs in the alternatives. Alpha is a probability of the best case occurring, and is set by the decision-maker; thus, if alpha is set wrong this method will give false results. This method generally gives a more balanced result than minimax or maximax.
4. Equally likely (Laplace) — Find the average payoff for each alternative, and choose the alternative with the highest average. The result will be the same as the CR technique with alpha set to .5, given only two possible states, and assumes that the best case and worst case scenarios are equally likely to occur.
5. Minimax Regret — Find the alternative that minimizes the maximum opportunity cost. Minimax Regret decisions are designed to produce acceptable results in all situations, by minimizing the negative consequences of being wrong.

Expected Monetary Value is the long-run average value of a particular decision. To develop an EMV, one must be able to assess the probability of each alternative occurring. After determining the expected monetary values of various alternatives, you pick the alternative with the highest EMV.

The Expected Value of Perfect Information (EVPI) places an upper bound on what one would pay for additional information. The EVPI is the expected value with perfect information (EV|PI) minus the maximum Expected Monetary value.

Five Steps to Decision Tree Analysis

1. Define the problem
2. Structure or draw the decision tree
3. Assign probabilities to the states of nature
4. Estimate payoffs for each combination of alternatives and states of nature
5. Solve the problem by computing expected monetary values (EMVs) for each state of nature node. Select the alternative with the best EMV.

Chapter 4:  Regression Analysis

Regression analysis is used to understand the relationship between variables and to predict the value of one variable based on another variable.

Regression models are used to test if a relationship exists between variables; that is, to use one variable to predict another.  There is normally some random error that cannot be predicted.

Y = Beta(zero) + Beta(1)X + error

Sample data are used to estimate the actual values for the slope and the y-intercept of the line generated.  The difference between the actual value of Y and the predicted value is known as the error.  Least squares regression minimizes the sum of the squared errors.