Advanced Decision Analysis
Nur Aini Masruroh
Sharpening the saw
Suppose you were to come upon someone in the woods working feverishly to saw down a tree.
• "What are you doing?" you ask.
• "Can't you see?" comes the impatient reply. "I'm sawing down this tree."
• "You look exhausted!" you exclaim. "How long have you been at it?"
• "Over five hours," he returns, "and I'm beat! This is hard work."
• "Well why don't you take a break for a few minutes and sharpen that saw?" you inquire. "I'm sure it would go a lot faster."
• "I don't have time to sharpen the saw," the man says emphatically.
"I'm too busy sawing!“
Scenario 1
Roy has prostate cancer. He has three alternatives: surgery, chemotherapy and playing a lot of golf. Prostate cancer is a very bad disease, so he will die soon, regardless of what he does.
He likes to play golf. If he chooses golf (and has no medical treatment), he will live on year in pain, but be able to enjoy his golf game during that time, then he will die.
If he undergoes chemotherapy, he will suffer tremendous nausea for six months and then he will either live for one good year or die right away.
If he has prostate surgery he will either live two good years or die on the operating table.
He says he wants to make his decision based on the probabilities assigned by his
doctor, who says that there is a 60% chance that the surgery will give its better outcome and a 60% chance that chemotherapy will give its better outcome. Which alternative should Roy choose?
Decision Making Under Uncertainties
Scenario 2
• Ms Anna, an attractive lady with graduate degree, must decide among many suitors. Among them include:
• Mr. A is very competitive person, holds a high paying job, but not religiously devoted.
• Mr. B loves to do community work during weekends, and holds a reasonable job with a stable income.
• Mr. C does not have a full time job but he is very active in religion related activities
Multiple Criteria Decision Making
Scenario 3
A company is certain that its market share for a product for the coming year will be between $0 million and $30 million.
The company would like to decide if it should mount a small or large advertising campaign during the coming year.
The company believes that its main rival , Fruit Inc., will mount either a small or large TV advertising campaign with a 50-50 chance.
The resulting market share and profits ($million) for the company depending on Fruits action are shown in the table.
What is the company’s best decision?
The company
chooses Fruit Inc chooses Small ad
campaign Large ad campaign Small ad
campaign 25%, $16 15%, $12 Large ad
campaign 35%, $8 25%, $10
Multiple Attribute Decision Analysis
Scenario 4
• Suppose university boards want to decide the structure of the curricula used to teach their students. The boards consist of lecturers, alumni, users, and benchmarking.
• How to formulate the curricula considering all the decision makers needs?
Group Decision Making
Why are decisions hard to make?
• Complexity
• There are many alternatives or possible solutions
• There are many factors to be considered and many of these factors are interdependent
• Uncertainty
• The possible future outcomes are uncertain or difficult to predict
• Information may be vague, incomplete, or unavailable
• Multiple conflicting objectives
• The decision maker(s) may have many goals and objectives
• Many of these goals or objectives may be conflicting in nature The decision makers are faced with the followings:
Why does DA offer?
• A conceptual framework for thinking systematically about hard and complex decision problems as to achieve clarity of action
• DA provides a set of tools for
• Structuring and representing the dependencies among all the alternatives and factors in the problem using decision trees and influence diagram
• Dealing with risks and uncertainty explicitly using probabilities
• Taking into account the risk taking attitude of the decision maker(s)
• Computing best alternatives under different future scenarios
• Managing information collection process for model enhancement
• Dealing with decision with multiple criteria and multiple attributes
• Dealing with group decision making
What is a decision?
• A decision is a commitment to a specific course of action that irrevocably allocates valuable resources (Ronald Howard)
• Example of decision:
• To fund a research project
• To take pursue this degree
• To go on an overseas holiday
• Example of non-decision:
• Regret not buying a certain item because the price has risen
• Worry about the not able to graduate from this program
Approaches to studies on decision making
• Normative approach: concerned with how rational decisions ought to be made
• Optimality
• Rationality
• Economics
• Descriptive (behavioral) approach: concerned with understanding how humans actually make decisions
• Developing psychological models of human cognition and thinking
• Explaining human behaviors
• Predicting human behaviors
Which approach to follow?
• Normative approach: ideal for complex decision situation as it ensures that rational decisions are made
• However psychological experiments have found that humans do not follow the norms prescribed by the normative approach
• They also found that humans are not very good in estimating probabilities
• Hence, normative approach alone is insufficient for real world applications
Prescriptive approach
• Prescriptive approach is normative approach taking into account descriptive
• It subscribes fundamentally to the normative approach and prescribe the procedures for decision modeling and analysis that take into
account the limitation of humans as revealed by the descriptive approach
• DA is based on prescriptive approach
Good decision versus good outcome
• Good decision is not guarantee good outcome – it only enhances the chance
Good decision
Bad decision Good
outcome Bad
outcome
Different roles in conducting DA
• Decision makers
• Stake holders or people who will make the final decision and will be held accountable for it
• Decision analyst or consultants:
• Person or team who facilitates the decision making process
• This course will train you to be a decision analyst or consultant
• The analyst must not assume the role of the decision maker (unless he is also the decision maker)
• Domain expert
• These are the people called in to provide expert opinion, expert judgment, etc
Elements of decision problems (Clemen and Reilly, 2001)
• Value and objectives
• Decision to make
• Uncertain events
• Consequences
Mention the decision elements of the previous examples!
Decision Analysis Process
• Howard, R.A, (1988). Decision Analysis: Practice and Promise. Management Science, 34(6), pp. 679 – 695
Formulation phase
• What is the decision?
• Four questions to answer:
• Which problems will be address in ths analysis? – scope of analysis
• What choices do we have? – the alternatives
• What make the choices difficult? – the uncertainties that affect the outcome of the decision
• What are the criteria for making a choice? – decision maker’s preferences
• The result is that we conceptualize and structure the decision problem into a model which also called the decision basis
Evaluation phase
• Address the question: “what is the recommended alternative?”
Appraisal phase
• Answer the question: “Why the recommended alternative should be implemented?”
• Sensitivity analysis is performed to evaluate the robustness of the recommendation
• Value of information analysis is performed
Sequential decisions
• Clemen, R.T. and Reilly, T. (2001). Making Hard Decisions with Decision Tools. California: Duxbury Thomson LearningThe coin tossing game or deal
• A coin is to be tossed. If you call the outcome (e.g. picture or number) correctly, you will be paid Rp 100.000, otherwise nothing.
• You have to pay some money to play the game
• The game is offered to you only once regardless of the outcome
• What is the maximum you are willing to pay for the right to play the game?
Personal Indifferent Buying Price
• The maximum you are willing to pay to play the game is called your personal indifferent buying price (PIBP)
• How did you arrive at your buying price?
• Did you perform any calculation to arrive it? What is it?
• Did everybody in the class have the same buying price?
Personal Indifferent Selling Price
• Suppose you bid for Rp X to won the right to play the previous game. Note that Rp X should be less then or equal to your personal indifferent buying price
• You pay Rp X and now has the right to play the game
• You can play the game immediately and resolve the outcomes, but you also have option to sell the right to another person and ask your friends to bid for it.
• What is the minimum amount for which you will sell the right to play the game to another person?
Personal Indifferent Selling Price
• The minimum you are willing to sell the game to another person is called your personal indifferent selling price (PISP)
• Your PISP is also called the certainty equivalent
• Is your PISP above or below Rp 50.000?
• What is so special about Rp 50.000?
• What can you say about your risk attitude?
• Is your PISP equal to your original PIBP for the deal?
Value of information
• While you are still thinking about selling the deal and is pondering about your uncertain future prospects on the deal a person came along and offer to tell you the outcome of the toss
• Suppose this person knows the outcome and will not lie to you, what is the amount you should be willing to pay him for the information?
• The above mount is called the value of information
Higher stakes: does everything scale up?
• The game is now played for Rp 500.000 or nothing
• What is your personal indifferent buying price?
• Is it 5 times that of the Rp 100.000 case?
• Suppose you win the right to the game. What is now your personal indifferent selling price?
• Is it 5 times that of the Rp 100.000 case?
The sunk cost principle
• Review of a concept from engineering economy
• You paid Rp 50.000 for a movie and halway through, found it pretty boring
• What should you do?
• Stay on no matter what since you have already paid Rp 50.000
• Sleep in the cinema hall and enjoy the air-con
• Walk out and do something more satisfying
• What you have already paid is called a sunk cost. It should not be considered in choosing the best alternative
Course outlines
Introduction to decision analysis, decision making tools Probabilistic thinking
Decision Theory and basic decision analysis Value of Information Analysis
Modeling Uncertainties using Bayesian Network Modeling Decision Using Influence Diagram
Modeling Risk Preference
Assessment of probability distribution Multiple Criteria Decision Making Multiple Attribute Decision Analysis
Group Decision Making
References
• Clemen, R.T. and Reilly, T. (2001). Making Hard Decisions with Decision Tools.
California: Duxbury Thomson Learning
• Howard, R.A, (1988). Decision Analysis: Practice and Promise. Management Science, 34(6), pp. 679 – 695.
• Russell, S. and Norvig, P. (2003). Artificial intelligent: A modern approach, 2 ed.
Prentice-Hall, Inc.
• Saaty, T.L. (1990). How to Make a Decision: the Analytic Hierarchy Process. European Journal of Operation Research, 48, pp. 9 – 26.
Readings
• Clemen, R.T. and Reilly, T. (2001). Making Hard Decisions with Decision Tools. California: Duxbury Thomson Learning (Chapter 1 and Chapter 2 )
• Howard, R.A, (1988). Decision Analysis: Practice and Promise.
Management Science, 34(6), pp. 679 – 695.