lecture 01 course overview - purdue university · 2020. 3. 16. · the need for studying decisions...

The Need for Studying DecisionsDecision Making: Normative and Descriptive Theories

ApplicationsCourse Structure

Lecture 01Course Overview

Jitesh H. Panchal

ME 597: Decision Making for Engineering Systems Design

Design Engineering Lab @ Purdue (DELP)School of Mechanical Engineering

Purdue University, West Lafayette, INhttp://engineering.purdue.edu/delp

August 22, 2019ME 597: Fall 2019 Lecture 01 1 / 49

http://engineering.purdue.edu/delp



Introduction

Jitesh H. PanchalAssociate Professor

School of Mechanical EngineeringPurdue University

At Purdue since Fall 2012

Assistant Professor at Washington State University (2008-2012)

Visiting Assistant Professor at Georgia Tech (2006-2008)Education

Ph.D. in Mechanical Engineering, Georgia Tech (2005)M.S. in Mechanical Engineering, Georgia Tech (2003)B.Tech. in Mechanical Engineering, Indian Institute of Technology, Guwahati(2000)

ME 597: Fall 2019 Lecture 01 2 / 49



Lecture Outline

1 The Need for Studying DecisionsExamples of decisions in designFlaws in decision making methods

2 Decision Making: Normative and Descriptive TheoriesHow SHOULD decisions be made?How do humans ACTUALLY make decisions?

3 ApplicationsProduct DesignSystems Design

4 Course StructureCourse contentsGrading scheme and Assignment 1

ME 597: Fall 2019 Lecture 01 3 / 49



Examples of decisions in designFlaws in decision making methods

What is Design?

Designing is the process of converting information about needs andrequirements for a product/system into a complete specification of thatproduct/system.

Role of decisions in design:

Design is to a large extent decision making.

Design involves deciding on the most preferred system alternative.

Can you think of decisions made in the design process?

ME 597: Fall 2019 Lecture 01 4 / 49




Who are the decision makers?

Decisions are everywhere...

1 Decisions made by customers in choosing products to purchaseProduct selection...

2 Designers as decision makersConcept selection...

3 Responding to decisions made by competitorsPricing...

4 Others?

ME 597: Fall 2019 Lecture 01 5 / 49




Example – Decisions Made by Customers

ME 597: Fall 2019 Lecture 01 6 / 49




Example – Decisions are Made Throughout a Product/System Lifecycle

Pilot Production

Production

Feasibility and

Concept

Detailed Design

Maintenance

& Support

Sales &

Distribution

Production

Testing

Detail

Design

Concept

Design

Portfolio

Planning

Product Lifecycle

Design

Reviews

Gate

Reviews

Concept Design

Review

Preliminary

Design ReviewFinal Design

Review

Production

Readiness ReviewProgram

Close-out

Program

Approval

Design Phase

Approval

Pilot Phase

Approval

Slide courtesy: Chris Paredis

ME 597: Fall 2019 Lecture 01 7 / 49




Example – Decisions Made by Designers

Conceptual design – morphological matrix

ME 597: Fall 2019 Lecture 01 8 / 49




Example – Decisions Made by Designers and Manufacturers

ME 597: Fall 2019 Lecture 01 9 / 49




Example – Decisions in Autonomous Transportation Systems

J.F. Bonnefon, A. Shariff, I. Rahwan, 2016, The Social Dilemma of AutonomousVehicles, Science, Vol. 352, No. 6293, pp. 1573-1576.

ME 597: Fall 2019 Lecture 01 10 / 49




Why do we need to study decision making?

Because poor decisions increase the likelihood of poor outcomes!

Poor decisions can result from:1 Flawed decision making methods2 Human biases

ME 597: Fall 2019 Lecture 01 11 / 49




Example 1: Voice of the Customer

Question

What is your preferred choice of product color? A,B,C,D, or E

A = Black

B = Blue

C = White

D = Red

E = Pink

Survey Results (100 customers):45 prefer A

25 prefer B

17 prefer C

13 prefer D

No one prefers E

Inference from the Survey: A > B > C > D > E

Hazelrigg, G. A., 2003, Validation of Engineering Design Alternative Selection Methods,Engineering Optimization, Vol. 35, No. 2, pp. 103-120.

ME 597: Fall 2019 Lecture 01 12 / 49




Preference Ordering

If designer considers preference ordering instead of just the top alternative,the following results are seen:

45 customers: A > E > D > C > B

25 customers: B > E > D > C > A

17 customers: C > E > D > B > A

13 customers: D > E > C > B > A

Verify that this preference structure will give the same results as seen onprevious slide.

In fact, there are 120 such preference orderings that will give same survey results.ME 597: Fall 2019 Lecture 01 13 / 49




Preference Ordering (contd.)

1 Count the number of customers that prefer A to E... = ???

2 Count the number of customers that prefer E to A... = ???

Perform similar pair wise comparisons for each pair.

What is the customers’ ACTUAL preference?

ME 597: Fall 2019 Lecture 01 14 / 49




Preference Ordering (contd.)

1 Number of customers that prefer A to E... = 45

2 Number of customers that prefer E to A... = 55

By performing similar pair wise comparisons, it can be seen that thecustomers’ actual preference is: E > D > C > B > A, which is exactlyopposite of the survey results!!!

Conclusion

The method is not reliable!!!

It is also possible that there is no alternative preferred by the group.ME 597: Fall 2019 Lecture 01 15 / 49



How SHOULD decisions be made?How do humans ACTUALLY make decisions?

The Process of Making a Decision

So, how SHOULD you make decisions?Is there a correct method?

ME 597: Fall 2019 Lecture 01 16 / 49




Favorable Properties of a Selection Method

1 The method should not impose preferences on the designer, that is,the alternatives should be ranked in accordance with the preferences ofthe designer.

2 The method should permit the comparison of design alternatives underconditions of uncertainty and with risky outcomes, including variabilityin manufacture, materials, etc., which pervade all of engineering design.

3 If the method recommends design alternative A when compared to theset of alternatives S{B,C,D, . . . }, then it should also recommend Awhen compared to any reduced set SR , such as {C,D, . . . } or{B,D, . . . } or {D, . . . }, etc.

Hazelrigg, G. A., 2003, ”Validation of Engineering Design Alternative Selection Methods,”Engineering Optimization, Vol. 35, No. 2, pp. 103-120.

ME 597: Fall 2019 Lecture 01 17 / 49




Favorable Properties of a Selection Method (contd.)

4 The method should make the same recommendation regardless of theorder in which the design alternatives are considered.

5 The method should be such that the addition of a new alternativeshould not make existing alternatives appear less favorable.

6 The method should be such that obtaining clairvoyance on anyuncertainty with respect to any alternative must not make the decisionsituation less attractive (information is always beneficial).

7 The method should be self-consistent and logical, that is, it should notcontradict itself and it should make maximum use of availableinformation for design alternative selection.

Hazelrigg, G. A., 2003, ”Validation of Engineering Design Alternative Selection Methods,”Engineering Optimization, Vol. 35, No. 2, pp. 103-120.

ME 597: Fall 2019 Lecture 01 18 / 49




Basic Elements of a Decision

Decision

A1

A2

An

O11

O12

O1k

O21

O22

O2k

On1

On2

Onk

U(O11)

U(O12)

U(O1k)

U(O21)

U(O22)

U(O2k)

U(On1)

U(On2)

U(Onk)

Select Ai

p11

p1k

p21

p1k

pn1

pnk

Alternatives Outcomes Preferences Choice

ME 597: Fall 2019 Lecture 01 19 / 49




Utility-theory for decision making

[vonNeumann and Morgenstern]

IF an appropriate utility is assigned to each possible consequence,AND the expected utility of each alternative is calculated,

THEN the best course of action is the alternative with the highest expectedutility.

ME 597: Fall 2019 Lecture 01 20 / 49




Behavioral Axioms

Behavioral axioms that form the basis of expected utility:Ordering of alternatives: Decision maker can order any twoalternatives and the ordering is transitive.Transitivity : If A1 � A2,A2 � A3, then A1 � A3

Reduction of compound uncertain events using complicated mixturesof lotteries using probability theory.

〈x∗, 0.2, x0〉 ∼ ???

〈〈x1, 0.2, x2〉, 0.5, 〈x3, 0.5, x4〉〉 ∼ 〈 ??? , 0.2, x0〉

〈〈(x1, y1), 0.2, (x2, y2)〉, 0.5, 〈(x3, y3), 0.5, (x4, y4)〉〉 ∼ 〈 (???, ???) , 0.2, (x0, y0)〉

Substitutability: reverse of reduction (i.e., the decision maker is willingto substitute a deterministic outcome with a lottery).

Clemen, R. T. (1996). Making Hard Decisions: An Introduction to Decision Analysis. Belmont,CA, Wadsworth Publishing Company. Chapter 14.

ME 597: Fall 2019 Lecture 01 21 / 49




Behavioral Axioms (contd.)

Continuity: If A1 � A � A2, then we can construct a lottery between A1

and A2 with probability 0 ≤ p ≤ 1 so that the decision maker isindifferent between A and the lottery.

Monotonicity: Decision maker prefers the gamble which has higherprobability of achieving the preferred outcome.

Invariance: All that is needed to determine a decision maker’spreferences among uncertain events are the payoffs and the associatedprobabilities.

Finiteness: No consequences are infinitely good or infinitely bad.

Clemen, R. T. (1996). Making Hard Decisions: An Introduction to Decision Analysis. Belmont,CA, Wadsworth Publishing Company. Chapter 14.

ME 597: Fall 2019 Lecture 01 22 / 49




Do people follow this normative model?

Do people really satisfy these behavioral axioms?Are people rational (as defined)?

ME 597: Fall 2019 Lecture 01 23 / 49




Tversky and Kahneman

Kahneman: 2002 Nobel Memorial Prize in Economic Sciences

“People reply on a limited number of heuristic principles which reduce thecomplex task of assigning probabilities and predicting values to simplerjudgmental operations.”

Examples of heuristics that people use in assessing probabilities:

1 Representativeness2 Availability3 Adjustment and anchoring

The use of these heuristics can lead to systematic errors!

Tversky, A. and Kahneman, D., 1974, “Judgment under Uncertainty: Heuristics and Biases,”Science, 185(4157), pp. 1124-1131.

ME 597: Fall 2019 Lecture 01 24 / 49




Representativeness:Estimating Probability from Similarity

“Steve is a very shy and withdrawn, invariably helpful, but with little interest inpeople, or in the world of reality. A meek and tidy soul, he has a need fororder and structure, and a passion for detail.”

Assess the probability that Steve is engaged in a particular occupation:

farmer

salesman

airline pilot

librarian

physician

ME 597: Fall 2019 Lecture 01 25 / 49




Representativeness:Insensitivity to prior probability of outcomes

Consider a school in a research university with 70 graduate students and 30faculty members.

John is a 30 year old man. He is married with no children. A man of highability and high motivation, he promises to be quite successful in his field. Heis well liked by his colleagues.

Assess the probability that John is a faculty member.

ME 597: Fall 2019 Lecture 01 26 / 49




Representativeness:Misconceptions of chance

In a sequential coin toss, which sequence is more likely?

“H-T-H-T-T-H”or

“H-H-H-T-T-T”or

“H-H-H-H-T-H”

Expectation that a sequence of events generated by a random process will represent theessential characteristics of that process even when the sequence is short.

ME 597: Fall 2019 Lecture 01 27 / 49




Representativeness:Misconceptions of chance

Gambler’s Fallacy: Chance is viewed as a self-correcting process in whichdeviations in one direction induces a deviation in the opposite direction torestore the equilibrium.

In reality, deviations are not “corrected”, they are merely diluted.

ME 597: Fall 2019 Lecture 01 28 / 49




2. Availability

Situations in which people assess the frequency of a class or the probabilityof an event by the ease with which instances or occurrences can be broughtto mind.

Are there more words in the English language that start with ‘r’ than wordswith ‘r’ as the third letter?

ME 597: Fall 2019 Lecture 01 29 / 49




In Summary

1 Most decision support tools in design and systems engineering (e.g.,QFD, Six Sigma, Pugh, Robust Design, Axiomatic Design, etc.) violatethe basic principles of decision theory. These methods are flawed, or atbest, rough approximations of normative decision theory.

Therefore, we must understand what the approximations are, and when thesemethods can be used safely.

2 Humans typically do not follow the basic principles of decision theory.They exhibit biases and follow heuristics for making decisions.

Therefore, we must understand how humans deviate from the ideal decisionmaking behaviors.

ME 597: Fall 2019 Lecture 01 30 / 49



Product DesignSystems Design

Applications

Applications in Engineering Design and Systems Engineering (EDSE)

1 Product design2 Systems design

ME 597: Fall 2019 Lecture 01 31 / 49




Multi-level Design

ME 597: Fall 2019 Lecture 01 32 / 49




Decisions in Modeling and Simulation

ME 597: Fall 2019 Lecture 01 33 / 49




Decisions Made by Analysts (contd.)

ME 597: Fall 2019 Lecture 01 34 / 49




Decisions Made by Participants in Open Innovation

ME 597: Fall 2019 Lecture 01 35 / 49




Failures in Space Systems

ME 597: Fall 2019 Lecture 01 36 / 49




Decisions Made by Participants in Power Systems

ME 597: Fall 2019 Lecture 01 37 / 49




Decisions Made by Participants in Air Transportation System

ME 597: Fall 2019 Lecture 01 38 / 49



Course contentsGrading scheme and Assignment 1

Introduction to ME 597

So, what is this course about?

ME 597 – Decision Making for Engineering Systems Design

Catalog Data: Multi-objective decision making under uncertainty;multi-attribute utility theory; applications to engineering design;Game-theoretic models with applications in engineering systems design;design for market systems; optimization-based formulation of Nash equilibria;applications to sustainable design and policy.

ME 597: Fall 2019 Lecture 01 39 / 49




Course Objectives

Upon completion of this course, students should be able to:1 Frame decisions and model designers’ preferences under risk2 Apply formal decision-making approaches to engineering design and

systems engineering3 Critically evaluate the assumptions and limitations of design decision

making approaches4 Formulate and execute decisions in presence of other decision makers

using game theoretic models5 Apply game-theoretic models to engineering systems design6 Adopt an interdisciplinary approach among engineering, economics, and

social sciences

ME 597: Fall 2019 Lecture 01 40 / 49




Overview of Topics

Context (2 weeks)Course overviewRole of decision making in Engineering Systems Design (ESD)Limitations of some popular decision making methodsElements of a decision problem

Normative Decision Theory (3 weeks)Tradeoffs and decision making under certaintyModeling uncertaintyMulti-attribute Utility theory

Applications in Engineering Systems Design (3 weeks)Utility-based selection in designDemand modelingSimulation-based design and experimentation

ME 597: Fall 2019 Lecture 01 41 / 49




Overview of Topics (continued)

Descriptive Decision Theory (2 weeks)Deviations from rationalityDescriptive decision theories

Multiple decision makers (1 week)Group preferences and aggregation of preferencesCollective decision making

Interactive Decisions with Applications (4 weeks)Strategic and extensive form gamesNash equilibriumApplications in ESD: Design for market systems; Negotiation protocols forsystems design; Open source product development; CrowdsourcingOverview of mechanism design

ME 597: Fall 2019 Lecture 01 42 / 49




Grading Scheme

Assignments 30%Mid-term Project 30%Final Project 40%

ME 597: Fall 2019 Lecture 01 43 / 49




Assignment 1

1 Basic information about yourself (name, employer, role, thesis advisor,research area, etc.)

2 Pedagogical goalsWhat were you expecting to learn when you registered for this course?Describe how this course relates to your longer-term goals.Having participated in the first lecture what are the three main learningobjectives that you will strive to achieve in this course?

Submission:1 Assignment 1 must be submitted online on Blackboard. Due Date:

Monday, August 26, 2019

ME 597: Fall 2019 Lecture 01 44 / 49




Key References

1 R. T. Clemen, 1996, Making Hard Decisions: An Introduction to DecisionAnalysis. Belmont, CA: Wadsworth Publishing Company.

2 R. L. Keeney and H. Raiffa, 1976, Decisions with Multiple Objectives:Preferences and Value Tradeoffs. New York: John Wiley and Sons.

3 R. Hastie and R.M. Dawes, 2010, Rational Choice in an UncertainWorld: The Psychology of Judgment and Decision Making, 2nd Edition,Sage Publications.

4 P.K. Dutta, 1999, Strategies and Games: Theory and Practice,Cambridge, MA: MIT Press.

ME 597: Fall 2019 Lecture 01 45 / 49




Additional References

1 G. A. Hazelrigg, 2012, Fundamentals of Decision Making ForEngineering Design and Systems Engineering:http://www.engineeringdecisionmaking.com.

2 R. A. Howard and A. E. Abbas, 2016, Foundations of Decision Analysis,Pearson Education.

3 R. L. Keeney, 1992, Value Focused Thinking. Cambridge, MA: HarvardUniversity Press.

4 K. E. Lewis, W. Chen, and L. C. Schmidt, eds., 2006, Decision Making inEngineering Design. New York, NY: ASME press.

5 W. Chen, C. Hoyle, H.J. Wassenaar, 2013, Decision-based Design:Integrating Consumer Preferences in Engineering Design, SpringerUSA.

6 R. B. Myerson, 1991, Game Theory: Analysis of Conflict. Cambridge,MA: Harvard University Press.

7 C. D. Aliprantis and S. K. Chakrabarti, 1999, Games and DecisionMaking. New York: Oxford University Press.

8 D. Fudenberg and J. Tirole, 1993, Game Theory. Cambridge, MA: MITPress.

ME 597: Fall 2019 Lecture 01 46 / 49

http://www.engineeringdecisionmaking.com




Emergency Preparedness – A Message from Purdue (1)

To report an emergency, call 911. To obtain updates regarding an ongoingemergency, sign up for Purdue Alert text messages, viewwww.purdue.edu//ea.

There are nearly 300 Emergency Telephones outdoors across campus and inparking garages that connect directly to the PUPD. If you feel threatened orneed help, push the button and you will be connected immediately.

If we hear a fire alarm during class we will immediately suspend class,evacuate the building, and proceed outdoors. Do not use the elevator.

ME 597: Fall 2019 Lecture 01 47 / 49

www.purdue.edu//ea




Emergency Preparedness – A Message from Purdue (2)

If we are notified during class of a Shelter in Place requirement for a tornadowarning, we will suspend class and shelter in [the basement].

If we are notified during class of a Shelter in Place requirement for ahazardous materials release, or a civil disturbance, including a shooting orother use of weapons, we will suspend class and shelter in the classroom,shutting the door and turning off the lights.

Please review the Emergency Preparedness website for additionalinformation.http://www.purdue.edu/ehps/emergency_preparedness/index.html

ME 597: Fall 2019 Lecture 01 48 / 49

http://www.purdue.edu/ehps/emergency_preparedness/index.html




THANK YOU!

ME 597: Fall 2019 Lecture 01 49 / 49

lecture 01 course overview - purdue university · 2020. 3. 16. · the need for studying decisions...

Documents