 |
Professor
Jonathon Caulkins,
Carnegie Mellon Heinz School |
Bio: Professor Jonathon Caulkins, a faculty member at Carnegie
Mellon's Heinz School since 1990, regularly teaches courses in Management
Science, Decision Analysis, Criminal Justice Policy, and Drug Policy.
He also conducts a Ph.D. seminar and occasionally advises project
courses. He has also taught at the RAND Graduate School and the
Technical University of Vienna.
Professor Caulkins was chosen as the 1999 winner of the prestigious
David R. Kershaw Award and its $10,000 honorarium by the Association
for Public Policy Analysis and Management (APPAM). He is the first
operations researcher/management scientist to win the Kershaw Award,
which recognizes individuals under the age of 40 who have made distinguished
contributions to the field of public policy analysis. The Kershaw
Award is given every two years, provided that a suitable recipient
is identified. It is named in honor of the first president of Mathematica
Policy Research, a nonprofit policy research organization. In addition,
Professor Caulkins won the Heinz School's Martcia Wade Award for
Teaching Excellence in 1999, and has been named a National Young
Investigator by the National Science Foundation.
Q: What classes are you teaching with Crystal
Ball?
Caulkins: Right now, I use Crystal Ball in the required
core Management Science course in Carnegie Mellon Heinz' Masters
of Science in Public Policy and Management. I plan to use it this
summer in a course at the Vienna Institute of Technology entitled
"Operations Research Modeling with Spreadsheets". Next
year, I will begin teaching a required core class called "Decision
Making Under Uncertainty" in our Masters of Information Systems
Management program.
Q: How are you teaching the software in your
classes?
Caulkins: I don't teach the software per se, because it
is sufficiently easy to use that the students can learn how to use
it from the textbook and/or just playing with it. Rather, I use
the software to teach management science and mathematical modeling
concepts, particularly for forecasting and simulation.
Q: How does Crystal Ball help your students?
Caulkins: Crystal Ball helps in two distinct ways. First
it makes the mechanics of things like simulation and time series
forecasting so easy that the class can focus on higher level conceptual
issues, such as whether the assumptions underlying the various methods
are met and what biases or error are likely to emerge when they
are met only imperfectly. In the past, so much time was spent teaching
the mechanics of how to produce the estimates that there was less
time for the more important and more subtle "forest level"
issues.
The second way that Crystal Ball helps is more fundamental. Many
people have trouble truly understanding random variables and the
associated distributions at an intuitive level. They pass the course,
but they never quite "get it" and so don't apply a "stochastic
mindset" to daily life.
Crystal Ball helps me teach that simply by making it so easy to
display and manipulate probability distributions graphically.
For example, I can easily show how changing the various distribution
parameters of a gamma distribution affects the shape of its probability
density function, and I can show how the central limit theorem takes
over and makes sums of random variables look approximately normal.
(e.g., the nth order negative binomial distribution becomes bell
shaped as the shape parameter gets larger.) Interactive, visual
descriptions of distributions are invaluable for teaching insight
into such probabilistic concepts.
Q: Why do you like the functionality of Crystal
Ball?
Caulkins: Two reasons: ease of use and generality. I can
teach textbook models that are elegant and apply in a very narrow
set of circumstances, but forecasting and simulation are simply
more often relevant. Almost all decision-making involves the future
and the future is always uncertain, so we are always forecasting.
And most stochastic models don't neatly fit into some form amenable
to closed for analysis (as the newsvendor problem does), so simulation
is the common, practical tool of choice for practicing professionals.
|
