Operations Research Chapter 3 Decision

Chapter III Decision-making
1. The concept
   of decision-making is aimed at decision-making problems with clear goals. After investigation and research, based on reality and possibility, a number of feasible schemes are drawn up, and
then a unified needle is used to select the whole process of pressing Party A.
2. Classification of decision-making
1. Different classification of decision-making methods: conventional decision-making and special decision-making.
2. Classified according to the relationship between planning and control: planned decision-making and controlled decision-making.
3. Decision-making procedure
1. Determine the goal
2. Draw up a feasible plan
3. Predict the natural state that may occur, calculate the income value of different plans in different states, compile the decision-making income table
4, and use the decision-making income table as the basis Different decision-making criteria are used for decision-making analysis and the optimal solution is selected.
Decision-making in different environments
Environmental classification
1.
Determining conditions: only one natural state 2. Uncertain conditions: respecting more than one natural state, the probability distribution is not fully known.
3. Risk conditions: there is more than one natural state. The probability distribution is totally unknown.
Decision-making under uncertain conditions
1. Maximum maximum decision criterion: The decision criterion of optimists, only considers the best natural state that will occur in the future, the probability of the best natural state occurrence is located 1, and other states are 0.
   Decision procedure: first Choose a maximum return value from each plan, and then select the largest return value from the plan, and choose the maximum value as an alternative plan. Take the big from the big
2. The maximum and minimum decision-making standard: the decision-making standard of conservatives. When making a decision, only the worst case is considered. The probability of the worst natural state is 1. The rest is 0
   Decision-making procedure: firstly select a minimum return value from each plan, and then choose the one with the largest return value as an alternative from the different plans represented by these smallest return values. Take the big from the small
3. The minimum and maximum regret value decision criterion:
Regret value: In the decision-making process, when a certain natural state may appear, the decision-maker did not choose the best option due to a decision error, but chose other options and regretted it. The difference between the return value of these two programs is the regret value.
   Decision-making procedure: first convert the income table into a regret value table, a regret value Wie basis, the larger one takes the smaller one.
4. Revealism decision-making standard: It is called this kind of decision-making standard. He sets the probability of the best pretend to appear in the future as α, and the worst state as 1-α. The other intermediate probabilities are 0, considering the two situations.
   The decision-making procedure is: Calculate the compromise value of each plan, and select the plan corresponding to the largest profit value after the compromise as the alternative plan.
Decision-making under risk conditions Decision-making under
     risk conditions is generally called statistical decision-making or random decision-making, which is mainly based on the probability that different natural states may occur, so the standard is mainly based on expected value labeling.
     When making decisions under risk conditions, the commonly used decision-making criteria are the expected profit criteria and the footer Bayes criteria. It mainly includes the maximum expected profit value standard and the minimum expected loss value standard. These two standards tend to be consistent with Russia in their rejection of the same question,
but the four decision-making standards under uncertainty are often inconsistent with Russia.
    Decision
tree The basic structure of the
decision tree is the decision point-the proposal branch ○-the probability of the state branch (not well adjusted here, don't believe it).