Informs 2015

62 / 552

INFORMS Philadelphia – 2015
60
SA63
63-Room 112B, CC
Nicholson Student Paper Competition
Cluster: Nicholson Student Paper Competition
Invited Session
Chair: Illya Hicks, Rice University, 6100 Main MS-134, Houston, TX,
77005, United States of America,
1 - Nicholson Student Paper Competition
Illya Hicks, Rice University, 6100 Main MS-134, Houston, TX,
77005, United States of America,
This session highlights the finalists for the 2015 George Nicholson Student Paper
Competition.
SA64
64-Room 113A, CC
Experts & Algorithms
Sponsor: Decision Analysis
Sponsored Session
Chair: Jason Merrick, Professor, Virginia Commonwealth University,
P.O. Box 843083, 1015 Floyd Avenue, Richmond, VA, 23284,
United States of America,
1 - Experts & Algorithms: Lessons in Blending Analytics with
Subjective Judgment
Cade Massey, Practice Professor, University of Pennsylvania, 3730
Walnut St., Huntsman 554, Philadelphia, PA, 19104, United
States of America, Jason Merrick
We share insights and tools gleaned from an 18-month project implementing an
analytics-based decision model for the admissions process at a large graduate
school. Issues include preference elicitation, forecasting, optimization and group
decision-making. We address both the theoretical and practical challenges
involved.
2 - Technology Implementation Decisions: A Multi-objective,
Multi- Stakeholder Case Study
Ed Cook, Capital One, , Jason Merrick
Major corporate technology adoption decisions are based on the financial case
and take an organization-wide viewpoint. However, technology implementation
decisions are different in nature and affect various stakeholders across the
organization with multiple, conflicting objectives. Our case study examines how
to implement a major technology change across a thousand branches of a
national bank and to avoid the costly delays and problems that afflicted other
banks implementing this technology.
3 - Overcoming Algorithm Aversion
Berkeley Dietvorst, Doctoral Student, The Wharton School, 3730
Walnut St, Suite 500, Philadelphia, PA, 19104, United States of
America, , Joe Simmons, Cade Massey
How can we get forecasters to use algorithms instead of human judgment? In four
experiments, participants decided whether to use an algorithm’s forecasts or their
own to complete a forecasting task. In the treatment conditions we gave
participants the option to modify the algorithm’s forecasts if they chose to use it.
Participants chose to use the algorithm much more often when they could modify
its forecasts, even when the amount that they could modify its forecasts was
severely restricted.
4 - Failure to Replicate Hyperbolic Discounting in
Large-scale Studies
Yael Grushka-Cockayne, Assistant Professor, Darden School of
Business, University ot Virginia, FOB 163,100 Darden Blvd,
Charlottesville, VA, United States of America,
Daniel Read, Casey Lichtendahl
We propose a simple model in which future income risk can explain hyperbolic
and anti-hyperbolic discounting. We present results from multiple experimental
studies involving thousands of participants, some of whom were asked Gallup-
type questions about future income. Overall, we fail to replicate hyperbolic
discounting. Instead, we find that anti-hyperbolic discounting effects are large and
reliable across studies, with little dependency on income expectations.
SA65
65-Room 113B, CC
Joint Session DAS/ENRE: Panel Discussion:
Climate Assessment and Decision Analysis
Sponsor: Decision Analysis & ENRE
Sponsored Session
Moderator: Melissa Kenney, Research Assistant Professor, University of
Maryland, 5825 University Research Court, Suite 4001, College Park,
MD, 20740, United States of America,
1 - Decision Analysis in the IPCC and National Climate Assessment
Robert Lempert, Senior Scientist, RAND, 1776 Main St, Santa
Monica, CA, 90407, United States of America,
Decision and risk analysis is playing an increasingly important role in scientific
assessments of climate change. This talk will describe this role, drawing on the
speaker’s experience as a lead author of the decision support chapters of both the
US National Climate Assessment and Intergovernmental Panel on Climate
Change (IPCC) Working Group II Fifth Assessment Report.
2 - Climate Change Decision Support Systems: Validation Required
Richard Moss, Senior Scientist, Pacific Northwest National Lab,
Joint Global Change Research Institute, 5825 University Research
Ct. Ste. 3500, College Park, MD, 20740, United States of America,
An increasing number of climate change decision support systems are being
offered for use. Overall, little is known about the effectiveness of many of these
tools. Failure to assess existing tools and practices is setting the research
community up for a loss of trust. This paper describes aspects of climate-related
decision support that require evaluation and argues that the decision analysis
community has a leadership role to play in this process through the US National
Climate Assessment.
3 - Climate Indicators: Do They Help or Hinder Decision Processes?
Melissa Kenney, Research Assistant Professor, University of
Maryland, 5825 University Research Court, Suite 4001, College
Park, MD, 20740, United States of America,
Though in Decision Analysis indicators measure objectives, in practice indicators
may not be linked to specific decision contexts. I will describe an effort to develop
recommendations to the U.S. Global Change Research Program for a wide-
reaching indicator system to inform decision about climate changes and impacts. I
will discuss preliminary work to determine if these kinds of indicators are useful
for decision structuring or as attributes in different decision processes.
4 - Discussant
Robert O’Connor,
5 - Risk and Resilience for Climate Change
Igor Linkov, Risk and Decision Science Team Lead, US Army
Corps of Engineers, Engineer R&D Center, 696 Virginia Rd,
Concord, MA, United States of America,
, Catherine Fox-lennt
An urgent need exists to complement the existing knowledge-base of climate
change risk by further developing frameworks enabling system-wide analysis. A
promising lens is resilience, a system property that can be planned for and
managed irrespective of adverse impact or system vulnerability. This presentation
will summarize our ideas and case study on the use of risk-based decision making
and decision-driven resilience management.
SA66
66-Room 113C, CC
Airlines Network Planning and Scheduling
Applications
Sponsor: Aviation Applications
Sponsored Session
Chair: Ahmed Abdelghany, Associate Professor, Embry-Riddle
Aeronautical University, 600 S. Clyde Morris blvd, Daytona Beach,
United States of America,
1 - Stochastic Fleeting with Itinerary Attractiveness in MapReduce
Diego Klabjan, Professor, Northwestern University, Evanston IL,
United States of America,
Stochastic fleeting models with discrete choice based on attractiveness are
computationally challenging. Parallel computing via Hadoop and MapReduce
have become ubiquitous. We study algorithms under the MapReduce parallel
framework for stochastic fleeting with attractiveness.
SA63