Decision Making 6

Question 1

Decision Theory Types

Answer 1

Normative:
- How decision makers make decisions
Philosophy, Economics, Game Theory

Descriptive:
- Explain and predict how people actually make decisions
Psychology, Behavioral Economics

Question 2

Challenges to rationality assumption

Answer 2

NY cabdriver problem

Question 3

Bounded Rationality

Answer 3

Rational models ignore situational and personal constraints such as time, pressure, and limited cog capacity

Using short-cut strategies such as heuristics

Question 4

Gambling Paradigm

Answer 4

elementary units of a decision are outcomes (consequences) and
probabilities: every decision can be reduced to a bet

Question 4

Utility Theory

Answer 4

Expected utility and marginal utility

Question 5

Prospect Theory

Answer 5

Utility, expressed in:
1. Decision weights:
impact that decision has on overall value of outcome
2. Value function:
reflects subjective value of outcome

Question 6

How do people arrive at judgements of probabilities and values?

Answer 6

1. Representativeness
   - Category Membership
   Probability of individual belonging to a category is related to category steroetype
   -> Heuristic attributes
   -> Ignoring base rates
   - Conjunction fallacy
   When people judge the conjunction of two events to be more probable then one of the events on its own
   - Random events
   - Gambler’s Fallacy
   When people think future independent events are affected by past events
2. Availability
   - Events are judged more likely to the extent that they are vivid or easily recalled
   -> Belief Bias
   Pre-exisiting beliefs influences evaluation of logical arguments
3. Anchoring and Adjustment
   - Estimation is off (by a lot) when anchors are given that are extreme and unrealistic

Question 7

MAUT

Answer 7

Multi Attribute Theory
- Find all alternatives
- Describe attributes of these alternatives
- Assess utility of each attribute value
- Rate importance of each attribute
- Select alternative with highest weighted value

1. Weighted adding
   - each attribute assigned value from 0 < v <1
2. Lexicographic Strategy
   - Based on most important attribute
3. Elimination by aspects
   - Eliminating options that do not reach a threshold

Question 8

Explicit VS Implicit Data

Answer 8

Explicit data is information that people voluntarily provide, implicit data is that which you must infer based on other data (watch time, streams, mouse movement, etc)

Question 9

User VS Item Based Collaborative Filtering

Answer 9

User based: find items that similar users liked
Item based: find similar items to ones that you already like

Question 10

Issues with Collaborative Filtering

Answer 10

1. Cold Start
   (new users have not made any reviews / new items have not been reviewed)
2. Sparsity
   each user only rates a few items, not accurate
3. Explainability
   how to explain recommendations?