Decision Analysis: Understanding Hard Decisions and Making Better Ones, Slides of Human-Computer Interaction Design

Download Decision Analysis: Understanding Hard Decisions and Making Better Ones and more Slides Human-Computer Interaction Design in PDF only on Docsity!

Introduction

2

Making Hard Decisions

 We occasionally need to make hard decisions

 What is the HARDEST decision that you have ever had to make?

 Decision Domains

 Personal domain  e.g. which career to pursue, where to live, etc  Business domain  e.g. which product to invest, where to locate  Government domain  e.g. how to cope with social problems, how to deal with international conflicts

4

Key Terms and Concepts (Cont’d)

 Uncertainty

 Something that is unknown or not perfectly known  e.g. Performance of stock market during the next five years

 Outcomes

 The possible things that can happen in the resolution of an uncertain event  e.g. The stock market is stable during the next five years

 Decision Context

 The particular decision situation which determines what objectives are considered  e.g. personal financial status, economic status of the nation  Decision context and objectives go hand in hand

5

Requisite Decision Models

 A model is considered requisite if contains everything that is

essential for solving the problem and only those that are essential

 It captures the essence of a decision modeling process

 It requires a full development of the decision maker’s thoughts

about the problem, beliefs regarding uncertainty, and preferences

7

Why Study Decision Analysis

 Decision Analysis

 A prescriptive approach designed for normally intelligent people who want to think hard and systematically about some important real problems  A tool to offer guidance to normal people making hard decisions based on fundamental principles and knowledge about human frailties in judgment and decision making

 Studying Decision Analysis Leads to Better Decisions

 Performance of decision making is better on average  Decisions are consistent  The same decision will be made given the same information  No surprises due to thorough study of the problem

8

Good Decisions

 What Is a Good Decision

 Emerges as a result of careful consideration of the available information and thorough deliberation about objectives and possible outcomes

 Good Decision  Good Outcome

 An outcome can be good because of good luck  Decision analysis cannot improve our luck, but it can certainly help us understand our problems better and thus make better decisions in general

10

Origins of Decision Analysis (Cont’d)

 Schlaifer’s (1959)

 Probability and Statistics for Business Decisions espoused Bayesian and decision analytic principles for business decisions

 Raiffa and Schlaifer’s (1961)

 Applied Statistical Decision Theory provided a detailed mathematical treatment of decision analysis, with focus on Bayesian Statistical Models

 Pratt (1964)

 Article “ Risk Aversion in the Small and in the Large ” made significant contributions to the theory of utility for money, formalizing a measure of risk aversion

 Howard (1966)

 First coined the term “decision analysis”

 Raiffa (1968)

 Decision Analysis established the decision analysis as a methodology in real applications

11

Process of Decision Analysis

• Objectives
  • e.g. min. cost, max. profit, improve health
• Alternatives
  • e.g. Invest or not invest, Choose A or B
• Decompose and Model
• “Divide & Conquer”
• modeling techniques, mathematical and statistical tools
• Sensitivity Analysis
  • Whether a slight change in one or more aspects of the model would affect the optimal decision

Such a process provides not only a structured way of thinking about decisions but also a structure in which the decision maker can develop beliefs and feelings

(Figure 1.1 in the textbook)

13

Objective:

Alternatives:

Decomposing and Modeling:

 Two possible outcomes: i) the airport is built at A, ii) the airport is built at B  Information available: the payoffs of all combinations of alternatives and outcomes, the probability of each outcome  Develop a decision tree model for the decision

 To maximize the profit

 Purchase a land at location A  Purchase a land at location B  Purchase a land at both locations A and B  Do nothing and wait till more information is obtained

14

Select the best alternative

 Purchasing a land at A has the highest expected payoff

Sensitive analysis

 The decision is not sensitive to small changes of model parameters

Implement the chosen plan of purchasing a land at A