Homework - Structuring Decisions
1) Explain the difference between End and Means Objectives.
2) What is an Objectives Hierarchy and what is it used to do?
3) How are objectives related to a decision's scope?
4) How are end consequences measured?
5) What is a Consequence Matrix? What information does it need to include?
6) Provide an influence diagram for the "decision in the presence of uncertainty" "pattern" that serves as the fundamental building block for all influence diagrams.
7) Describe a difficult decision that you have had to make (or will be making). Develop an annotated influence diagram for this decision that can serve as a requisite model. This should include decision options, outcome and outcome probabilities, and the potential end states (for each decision option).
8) Provide a decision tree for the "decision in the presence of uncertainty" "pattern" that serves as the fundamental building block for all influence diagrams.
9) Develop a decision tree for the decision described in 7.
10) Which method (influence diagram or decision tree) to you prefer? Why? Note: while you may discuss pros and cons here, I expect you to make a decision as to which method you prefer (which does not have to be the same as my preference).
11) What is the Clarity Test? What is it used to do?
12) How does the selection of poor attributes and attribute weights affect a decision?
13) In general can hard decisions be characterized by single directly measurable attribute? Explain.
14) Explain how one might develop a single metric to represent the desirability of an end state characterized by values for a set of N measurable attributes. What kind of analysis is this? What is this metric called?
Some Other Potential Test Questions:
1) Briefly describe each of these three methods for structuring objectives:
a. Objectives List
b. End Objectives/Values Hierarchy
a. Means Objectives Network Diagram
2) Briefly describe why you would want to construct each of these (i.e., what do they help you to do?).
3) What are the consequences of defining objectives too narrowly? Too broadly?
4) What are some tests for whether you have defined your objectives too broadly?
5) What are the two principal methods for structuring (schematically representing the architecture of) a decision or series of decisions?
6) How does an "annotated" influence diagram differ from an "unannotated" influence diagram? Which may serve as a "requisite model?" Explain.
7) How are Decisions, Chance Events, Consequences, and the relationships between them represented in an Influence Diagram? How does one annotate each of these?
8) What is a "Calculation Node?" How is it annotated?
9) What is the difference between a "consequence arc" and a "relevance arc?"
10) How does one build an Influence Diagram?
11) What does it mean to "elaborate" an influence diagram?
12) How does an Influence Diagram differ from a flow chart?
13) What may you conclude if a sequence of consequence and relevance arcs results in a cycle (i.e., leads one back to earlier decision or event)?
14) What does a chance node pointing to a decision node imply?
15) How are Decisions, Chance Events, Consequences, and the relationships between them represented in a Decision Tree?
16) Explain how Decision Trees "self-annotating?"
17) Explain how developing a Decision Tree goes a long way toward developing a "requisite model" for a decision (or series of decisions).
18) What two things need to be specified for each branch coming out of an Uncertain Event node?
19) What does it mean for an chance node's branches to be "mutually exclusive" and "collectively exhaustive?" Why is this important?
20) Explain what is meant by each of the following three different types of branching that can occur at an Uncertain Event node:
a. Discrete binary
b. Discrete multilevel
c. Continuous
21) If a decision tree does not capture a potential set of decisions, uncertain events, or end states, what may you conclude about it?
22) What is meant by a decision tree or influence diagram "pattern?"
23) What are three key decision tree/influence diagram patterns?
24) Provide an example of an attribute with quantitative values.
25) Provide an example of an attribute with qualitative values.