Decision-Making Under Uncertainty: Criteria and Tables

Hurwicz Criterion: Coefficient of Optimism

In this approach, a decision-maker may have an intermediate attitude. Since no probabilities are associated with each state of nature, this author suggests using a coefficient of optimism, simultaneously called C, and a coefficient of pessimism, (1 - C), where 0 ≤ C ≤ 1.

The coefficient C indicates the decision-maker's stance toward risk. When it is closer to 1, the decision-maker will be more optimistic, while closer to 0, the decision-maker is more pessimistic.

This criterion focuses only on the extreme scores of each alternative, weighing the result of maximum optimism with the coefficient C and the result of minimal value with the coefficient of pessimism (1 - C). The sum of these two products will provide the value of each alternative, and the one that offers the best value will be chosen.

Suppose the manager is very cautious and pessimistic, assigning their coefficient of optimism a value of C = 0.2.

Hurwicz = Maximax * C + Maximin * (1-C)

Disadvantages

Like the other two rules, this approach ignores the less extreme results.

Minimax Regret Criterion

This is the difference between the actual payment and the payment that would have been obtained if the optimal act had been chosen for the state of nature that was obtained.

Laplace Criterion

This consists of assigning the same probability of occurrence to each state of nature.

Expected Value Criterion

This approach uses all the information furnished by a payoff matrix. It is based on statistical probabilities. It is advisable to assign a priori different states of nature, using historical data, other available information, or simply your subjective opinion. We choose the most convenient.

Unscheduled Decisions

In cases of non-repetitive, unpredictable decisions with a high degree of uncertainty, use the decision matrix as a tool to facilitate the choice between different actions or alternatives. For its construction, the following questions are suggested:

Decision Table

These are a means of graphic expression of algorithms. They consist of the tabulation of all the different conditions or variables of the problem and the respective action to be taken for each different possible combination thereof. Both the conditions and actions must be evaluated.

The terms are broken down so that each presents only dichotomous values:

• Y = yes
• N = no
• - = Irrelevant (or unenforceable)

Actions can take the values:

• X = execute
• - = Do not execute