2016 INFORMS Annual Meeting Program

350 / 561

INFORMS Nashville – 2016
350
3 - Conic Optimization Model For Replicated Data Stores In Geo-
distributed Cloud Applications
Julio Cezar Goez, NHH Norwegian School of Economics, Bergen,
Norway, Juan F. Pérez
We consider a software application provider that serves geographically distributed
users using cloud resources. The application provides a service to access content
via a set of channels, and it must comply with a certain quality of service (QoS).
The provider must decide where to locate and how to replicate the data
considering traffic patterns. The goal is to find the deployment of minimum cost.
We introduce a mixed integer non-linear optimization model, which may be
reformulated as a mixed integer second order cone optimization problem. In
many of our test instances CPLEX reaches the time limit without feasibility. We
developed a feasibility test that also provides an initial feasible solution.
4 - Large Scale Dynamic Network Revenue Management With
Application In Pricing Of Interactive Cloud Applications
Hossein Jahandideh, UCLA Anderson School of Management,
3777 Mentone Avenue, Apt 405, Los Angeles, CA, 90034-6473,
United States, , Julie Ward, Filippo Balestrieri
We consider a cloud provider making customized dynamic pricing decisions for
hosting interactive applications. The dynamic pricing problem is a large-scale
dynamic network revenue management problem. We decompose the stochastic
dynamic program into single-resource problems by exploiting structural
properties of the optimal Lagrangian variables, removing the necessity for an
extensive search over Lagrangian variables. We define a pricing mechanism based
on the solution to the decomposed problem, and demonstrate its effectiveness
through several numerical experiments.
TD48
210-MCC
Social Media Analytics Award Session
Invited: Social Media Analytics
Invited Session
Chair: Tauhid Zaman, MIT, 77 Massachusetts Ave, Boston, MA, 0,
United States,
1 - Social Media Analytics Best Student Paper Award
Tauhid Zaman, MIT, Atlanta, GA,
Presentations by the finalists in the Social Media Analytics Best Student Paper
Award Contest.
TD49
211-MCC
Panel: First Course in Analytics or Only Course in
Analytics – What Difference Does It Make?
Sponsored: Education (INFORMED)
Moderator: Thomas G Groleau, Carthage College, Kenosha, WI, United
States,
1 - Exploratory Questions
Thomas G Groleau, Carthage College,
The moderator will explain the purpose of the panel, provide examples from
other disciplines, and propose a few questions to the panel to get the discussion
started.
2 - Panelist:
Peter C Bell, Ivey Business School at Western University, Canada,
3 - Panelist:
Jeffrey D Camm, Wake Forest University,
4 - Panelist:
Dimitris Bertsimas, Massachusetts Institute of Technology,
5 - Panelist:
Robert Krider, Professor, Simon Fraser University, Vancouver, BC,
Canada,
TD50
212-MCC
SpORts: Sports Analytics II
Sponsored: SpORts
Sponsored Session
Chair: Stephen Hill, University of North Carolina-Wilmington, 601
South College Road, Wilmington, NC, 28403-5611, United States,
1 - Breaking The Chill Of The Tie In The National Hockey League
Marty Thomas, Georgia Gwinnett College,
In order to increase fan excitement in overtime periods, the National Hockey
League changed the point structure of the overtime period (winner receives 2
points and the loser receives 1 point). However, tied games now had more points
associated with them (regulation win = 2 points; regulation loss = 0 points). This
research explores the implications of a soccer point system (3 points = regulation
win; 2 points = overtime win; 1 point = overtime loss). The Real Time Scoring
System is used to develop probability distributions of goal production for each of
the 30 NHL teams. Monte Carlo simulation is used to determine whether the new
point system benefits certain teams over others in terms of making the playoffs?
2 - The Advantage Of Lefties In Sports
Hal Cooper, PhD Candidate, Columbia University, 500 West 120th
Street, Rm. 345, Mudd, New York, NY, 10027, United States,
, Francois Fagan, Martin B Haugh
Left-handers comprise a staggering 15% of professional tennis players, but only
11% of the general population. In sports as varied as boxing, baseball and fencing,
the contrast is even more stark. Here we present a method for extracting the
advantage of being left-handed in sports (as well as the inherent skill of each
player) from match results. Unlike previous approaches to this problem, our
formulation is Bayesian and uses induced order statistics to address the truncated
nature of the data set. We further demonstrate an approach in the absence of
explicit match result data that can be used to determine the latent advantage of
specific factors wherever there exists notions of ranking and competition.
3 - The Optimal Value Bill James’ Pythagorean Method For Major
League Lacrosse
Hayden Howell, University of Alabama, Culverhouse College of
Commerce and Business Administration, Tuscaloosa, AL, 35487-
0226, United States, , James Cochran
Bill James’ Pythagorean Method of Baseball, which quantifies the nature of the
relationship between the win/loss percentage of a Major League Baseball (MLB)
team and the number of runs the team scores and allows over the course of a
season, is extended to Major League Lacrosse (MLL). We find the optimal form of
James’ model using both the squared and the absolute error criteria over a broad
range of algebraic possibilities. We also examine the stability in the relationship
between win/loss percentage and runs scored and allowed over time.
4 - Determining The Optimal Locations For Indoor Tennis Camps In
Canada
Islay Wright, University of Toronto, 1, Toronto, ON, Canada,
Timothy Chan
Winter tennis camps are often inaccessible for Canadians due to the high rental
costs of indoor tennis facilities. To increase the involvement of youth in tennis, it
is important to make camps available year-round at a reasonable price. Tennis
Canada wants to evaluate holding tennis camps at indoor turf stadiums instead of
at indoor tennis facilities. To help them, we developed a location model to
optimize camp locations, using demand estimates generated by a logistic
regression model and data from surveys and census.
TD51
213-MCC
Models of Influence and Optimal Response
Sponsored: Public Sector OR
Sponsored Session
Chair: Theodore T Allen, Ohio State University, 210 Baker Systems
Engineering, 1971 Neil Avenue,, Columbus, OH, 43210, United States,
1 - The Role Of Peer Influence On Vaccine Uptake In A Pandemic
Disease Spread Model
Kevin Chan, University of Toronto, Toronto, ON, Canada,
, Dionne Aleman
Most pandemic spread models that consider vaccination assume vaccine adopters
are evenly spread across the population. However, evidence suggests vaccine
adoption is peer-influenced. Treating populations as contact networks with
vaccination determined by influence-spread models, we analyze pandemic
outcomes using agent-based simulation. Far more infections occur with peer-
influenced v. uniformly-spread vaccination.
TD48