Week 5 & 6

Question 1

What are the steps of analyzing decision problems that involve uncertainty?

Answer 1

1. Identify the problem
2. Figure out possible decision
3. Figure out possible outcomes
4. Figure out the probability of each outcome
5. See what is the cost and benefits/payoff?
6. Check with decision criteria and make decision

Question 2

What are the decision tree components?

Answer 2

Decision tree are composed of nodes (the shapes) and branches (the lines):

A decision node (square): represents a choice to be made (cross off unwanted option)
An event node (circle): things that happen outside our control
A terminal node (triangle): indicates the end of the process, a completed problem

Question 3

How does time proceed/flow in the decision tree?

Answer 3

Left to right

Question 4

Where to place probabilities?

Answer 4

In the probabilities branches

Question 5

Where to place the payoffs?

Answer 5

The right side of the terminal nodes (triangle)

Question 6

What is an EMV?

Answer 6

Expected Monetary Value = weighted average of the possible payoffs/cost for the decision weighted by the probabilities of outcome
Choose the LARGEST EMV

Question 7

What is a one stage / single stage decision problem?

Answer 7

You only make one decision, the one right now; then you wait for an uncertain outcome

Question 8

What is a multi-stage decision problem?

Answer 8

You make a decision one step at a time, and see what happens, then use this new information as tool to make the next decision

Question 9

What is the objective of multi-stage decision problem?

Answer 9

To maximize EMV, known as contingency plan

Question 10

What is the rolling back procedure?

Answer 10

Systematic way of calculating EMV in the decision tree from the end of tree and works back to the beginning
At decision node: Choose highest EMV
At probability node: Calculate EMV

Question 11

Difference of prior vs posterior probability?

Answer 11

Prior is before new information is acquired, posterior is the probability of the event AFTER the new info is acquired

Question 12

What is p(A|B)

Answer 12

The probability that event A occurs given that event B is known to occur

Question 13

What is a simple Bayes’ Rule?

Answer 13

P(A|B) = P(B|A) x P(A) / P(B)

Question 14

Bayes’ Rule: How to find P(A)?

Answer 14

P(A) =
Example = P(B) = P(G (green box)) x P(B|G) + P(Y) x P(B|Y)

Question 15

What is p(Ai | B)?

Answer 15

The probability that B comes from Ai
Example: Prob that broken part (B) comes from supplier 2 (A2 -> Ai) instead of supplier 1 (A1)

Question 16

How do you write: Prob that broken part from supplier 2 is 2%?

Answer 16

p(B | A2) = 0.02 -> p(B | Ai)

Question 17

What is EVPI?

Answer 17

EVPI is Expected Value of Perfect Information, how much are you willing to pay for the perfect information
EPVI = EV with PI - EV without PI

Question 18

What is attitude towards risk?

Answer 18

How people behaviour is influence by potential risk, people are willing to spend on insurance so that expensive property is insured even though the EMV is -