Patrick Gurian's Webpage

Decision Analysis in Engineering and Science

Fall 2005
Wednesday 6:00-8:50 p.m.
Revised 10/13/2005

Professor:  Patrick Gurian
Office:  Alumni Engineering 270-K

Phone:  215-895-2889

Office Hours:  Wed. 3-4 pm, Thurs. 3-4 and by appointment

Text:  Making Hard Decisions with DecisionTools Suite, Clemen and Reilly

Course Overview

This class covers mathematical approaches and software tools for structuring and solving decision problems.  Students will learn techniques for making decisions under uncertainty, assessing trade-offs between different objectives, and determining the value of further information.  The students will learn to use five decision making software tools that are included with the text:  @RISK, PrecisionTree, TopRank, BestFit, and RISKview.  Applications to engineering will be emphasized, but the course material is generally relevant to decision making in public health, public policy, and business.

Course Policy
Students needing special accommodations for exams or coursework must submit letters in accordance with university policy.  Please turn off cell phones during class meetings.  Computers (any device with wireless connectivity or a full alpha-numeric keyboard) may not be used on exams.  Exams and quizzes can not be rescheduled because of prior travel arrangements.  If you have a conflict with either the final exam or midterm exam you should not enroll in this class.  Late assignments are penalized 3% per day late.  Homework assignments will routinely be announced by email and posted on WebCT.  Students are responsible for ensuring that they are on the class email list and for checking WebCT on a weekly basis.

Assignments (Grading)

Routine Problems (ungraded)
Students will be assigned problems after each lecture and answers will be posted on WebCT.  While these are not graded they are good preparation for quizzes.  They may also involve postings to the class message board (see below).

Virtual Class Participation (10%)
In some cases posting to the WebCT message board will be required.  If your posting receives less than full credit you will be notified by email.

Software Projects (15%)
Two small projects will be assigned which will consist partly of documenting your results from the in-class activity and partly of extending the analysis and integrating your results with team members.  A hard copy must be turned in to the instructor in person, not by email, not left in a mailbox. 

Quizzes (10%)
A quiz will typically be given at the start of class.  Each quiz will require about 10 minutes to complete.  The lowest quiz grade will be dropped.

Midterm (30%)
An exam will be given in class on October 26.

Final Exam (35%)

The final exam will be cumulative but will emphasize material covered after the mid-term.

Course Schedule




Sept. 28

Introduction, decision trees, risk profiles, stochastic dominance

Chapters 1-4

Oct. 5

Abbreviated class:  influence diagrams


Week of Oct. 5

Virtual class:  Review of discounting


Oct. 12

Sensitivity analysis, Conditional probability, Bayes’ Theorem

Chapter 5
Chapter 7

Oct. 19

Dicretizing, first order uncertainty analysis

Chapter 8

Oct. 26

Midterm (Nov. 4 withdrawal deadline)


Nov. 2

No class meeting
Virtual Class:  Developing a decision tree
Monte Carlo simulation
Office hours cancelled this week-please email your questions

Chapter 9
Chapter 11
Monte Carlo project

Nov. 9

Value of information
Risk preferences, utility functions, eliciting utility functions

Chapter 12
Chapter 13

Nov. 16

Multi-objective decision making, assessing trade-off weights, dominance, conflicting objectives

Chapter 15-16

Nov. 23

Thanksgiving:  No class


Nov. 30

Utility theory axioms, Allais paradox, divergences between normative and descriptive decision making

Chapter 14

Dec. 7

Subjective probability, heuristics and biases


Dec 14

Final Exam