Documentos de Académico
Documentos de Profesional
Documentos de Cultura
5. DECISION THEORY.
Clearly Define The Problem. List The Possible Alternatives. Indentify The Possible Outcomes. List The Payoff or profit of each combination of Alternatives & Outcomes. 5. Select one of The Mathematical Decision Theory Models. 6. Apply The Model and Make your Decision.
26
5.1.3.
5.1.4. DECISION MAKING UNDER RISK. Decision Making under risk is a probability decision situation. Several possible states of nature may occur, each with a given probability. In this section, we consider one of the most popular methods of making decisions under risk, namely, selecting the alternative with the highest expected monetary value. EXPECTED MONETARY VALUE (EMV). FORMULAE.
EMV(Alternative i) =
(Payoff(state
j) * Prob(X=xj)
200 100 0
0.5
-180 - 20 0
0.5
EMV 10 40 0
Select decision = A 2 DEFINITION. EVPI (Expected Value of Perfect Information) is defined as following: EVPI = Best Outcome for State i* P[X=xi] - EMV From the above example, we have: EVPI = 200/2 +0/2 - 40 = 60. Thus the most Thompson would be willing to pay for perfect information is 60$.
27
OPPORTUNITY LOSS. An alternative approach to maximizing expected monetary value (EMV) is to minimize EXPECTED OPPORTUNITY LOSS (EOL). Opportunity loss, sometimes called regret, refers to the difference between the optimal profit or payoff and the actual payoff received. METHOD FOR CALCULATING THE EOL. STEP 1. Create the Opportunity Loss Table (OL). FORMULAE. OL(Alter. j, State i) = Best Outcome of State i - Actual Outcomes STEP 2. Compute The EOL by the following Formulae: EOL(Alternative j) = OL(Alternative j,State i)* P[X=xj] For the above example, we have The OL Table as follows:
ALTERNATIVES STATES OF FAVORABLES MARKET NATURE UNFAVORABLE MARKET
200 200 =
0 = 200
EOL 90 60 100
Select decision = A 2 NOTE. Minimum EOL will always result in the same decision as Maximum EMV, and that The following relationship always hold: EVPI = Min EOL.
28
When the probability of occurrence of each state of nature cannot be assessed, we can use the following criteria for making the decision: Maximax. Maximin. Minimax. Equally likely. Criterion realism. 5.1.5.1. MAXIMAX. The Maximax Criterion finds the Alternative that maximizes the maximum outcomes or consequence for every alternative. For the above example, we have:
ALTERNATIVES STATES OF FAVORABLES MARKET NATURE UNFAVORABLE MARKET MAX in Row
200 100 0
0.5
-180 - 20 0
0.5
200 100 0
Select decision = A 1 5.1.5.2. MAXIMIN. The Maximin Criterion finds the Alternative that maximizes the minimum outcomes or consequence for every alternative. For the above example, we have:
ALTERNATIVES STATES OF FAVORABLES MARKET NATURE UNFAVORABLE MARKET MIN in Row
200 100 0
0.5
-180 - 20 0
0.5
-180 - 20 0
Select decision = A 3
29
5.1.5.3. MINMAX. The Mimax Criterion based on opportunity loss. Minmax finds the Alternative that minimizes the maximum opportunity loss within each alternative. For the above example, we have:
ALTERNATIVES STATES OF FAVORABLES MARKET NATURE UNFAVORABLE MARKET MAX in Row
0 100 200
0.5
180 20 0
0.5
200 100 0
0.5
- 180 20 0
0.5
10 40 0
Select decision = A 2
30
5.1.5.5. CRITERION REALISM (WEIGHTED AVERAGE = W.A). The criterion is a compromise between an optimistic and a pessimistic decision, with a coefficient of realism a. FORMULAE. W.A = a(Max in Row) + (1-a)(Min in Row) For the above example, we have:
ALTERNATIVES STATES OF FAVORABLES MARKET NATURE UNFAVORABLE MARKET W.A , a=.8
200 100 0
0.5
- 180 20 0
0.5
124. 76 0
Select decision = A 1
31
For The above example, we have: State of Nature Node Favorable Market EMV(1)=10 Plant
Large
5.2.2.
BAYES S FORMULAE
BAYES S FORMULAE
P P(A\B) = P( B \ A )P(( B \) A )P((BA\)A )P( A ) P A +
32
ALTERNATIVES A1. Large facility A2. Small facility A3. Do nothing Survey Positive (SP) Survey Negative (SN)
P(UM\Survey positve) =
P(FM\Survey negative) =
P(UM\Survey negative) =
P(Survey Result Positive) = P(SP\FM)P(FM) + P(SP\UM)P(UM) =0.45 P(Survey Result Negative)=P(SN\FM)P(FM) + P(SN\UM)P(UM) =0.55
33
5.3. EXERCISES.
1. Consider the following profit table, where I.. V are Alternatives, S1, ..,S5 are states of nature. The prior probabilities associated with S1 to S5 are .3, .2, .1, .15 and .25 respectively.
I II III IV V S1 10 17 05 35 45 S2 31 67 45 19 51 S3 11 13 13 09 08 S4 39 22 18 29 40 S5 54 43 32 35 45
Which Alternative maximizes expected monetary value? Which is the best decision if you use the opportunity loss. 2. Suppose the probabilities of occurrence of each state not be assessed. Which is the best decision if you use the following criteria: Maxmax, Maxmin, MinMax, Average. For the above Table. 3. The following payoff matrix indicates the monetary values that would be realized for each of the three alternatives (A 1, A 2, A 3 ) and three states of nature (S1, S2, S3).
A1 A2 A3 Prob, S1 1000 0 -400 .3 S2 -100 0 100 .5 S3 -2000 100 500 .2
3.1. Which Alternative maximizes expected monetary value? 3.2. A market analyst offers to conduct a survey to determine which state of nature migh occur. The result of survey as follows:
Result of Survey Positive Negative S1 .7 .3 S1 900 -100 -500 .3 S2 .2 .8 S2 -200 -100 0 .5 S3 .2 .8 S3 -2100 0 400 .2
A1 A2 A3 Prob,
Determine The Posterior probabilities. Draw a decision tree to represent these situations and find the best strategy.
34