Informs 2015

433 / 552

INFORMS Philadelphia – 2015
431
4 - Multi-stage Stochastic Lot Sizing Problem with
Nervousness Considerations
Esra Koca, Asst. Prof., Sabanci University, Department of
Industrial Engineering, Istanbul, 34956, Turkey,
M. Selim Akturk, Hande Yaman
We study the multi-stage stochastic lot sizing problem with controllable
processing times and nervousness considerations. In multi-stage stochastic
programming, one can obtain different production decisions for different
scenarios and this situation may lead to lack of coordination. We formulate the
problem by considering this drawback of the approach. Some mixing set
structures are observed as relaxations of our formulation, and valid inequalities
are developed based on these structures.
5 - Chance Constrained Optimization for Pari-mutuel
Horse Race Betting
Michael Metel, PhD Student, McMaster University, 1280 Main St.
West, Hamilton, ON, L8S4M4, Canada, Michael Metel,
, Kai Huang, Antoine Deza
We consider the time horizon of a gambler in the optimization of horse race
betting through the use of chance constrained programming. A novel approach to
estimating superfecta payouts is presented using maximum likelihood estimation.
A computational substantiation with historical race data found an increase in
return of over 10% using the chance constrained model.
WC13
13-Franklin 3, Marriott
Data Driven Optimization and Applications II
Sponsor: Optimization/Optimization Under Uncertainty
Sponsored Session
Chair: Jun-ya Gotoh, Chuo university, 1-13-27 Kasuga, Bunkyo-ku,
Tokyo, 112-8551, Japan,
1 - Performance Analysis of Stochastic Dynamic Programs via
Information Relaxation Duality
David Brown, Duke University Fuqua School of Business,
1 Towerview Rd, Durham, NC, United States of America,
, Santiago Balseiro
A common technique in the analysis of stochastic systems is the use of “hindsight
bounds,” in which decisions are made after uncertainties are revealed. In some
applications, however, hindsight bounds may lead to very weak performance
guarantees. We show how to obtain stronger guarantees by incorporating
penalties that punish the use of additional information, and demonstrate the
technique on several applications, including stochastic knapsack problems.
2 - A Unified Classification Algorithm Based on Accelerated Proximal
Gradient Methods
Akiko Takeda, The University of Tokyo, 7-3-1 Hongo, Bunkyo-
ku, Tokyo, 113-8656, Japan, ,
Naoki Ito, Kim Chuan Toh
The goal of binary classification is to predict the class of a new sample. It is
important to find suitable classification models for given datasets for good
prediction performances. We design a unified model for various types of
classification models and an efficient algorithm for solving the unified model,
which speeds up the process of finding the best model. It is based on an
accelerated proximal gradient method and performs better than specialized
algorithms for specific models.
3 - Demand Clustering and Inventory Optimization for
Fashion Products
Tong Wang, Associate Professor, National University of Singapore,
15 Kent Ridge Drive, Singapore, SG, 119245, Singapore,
We study demand estimation and inventory optimization for fashion products.
Based on historical product covariates and sales, we construct Bayesian structure
model to cluster products that share commonality in covariates and sales pattern.
At the beginning of a new season after covariates of new products are observed,
they are assigned into the clusters, and their demand distributions are estimated.
Inventory decisions are made accordingly. Clustering is updated at the end of the
season.
4 - Robust Empirical Optimization is Almost the Same as
Mean-variance Optimization
Jun-ya Gotoh, Chuo university, 1-13-27 Kasuga, Bunkyo-ku,
Tokyo, 112-8551, Japan,
Michael Kim, Andrew Lim
We consider a distributionally robust optimization (DRO) problem in which the
decision maker optimizes against a worst-case distribution, where a penalty on
phi—divergence controls the amount that the alternative can deviate from the
nominal. Our main finding is that robust empirical optimization is essentially
equivalent to solving an in-sample mean-variance problem, which provides
insight into the mechanism by which empirical DRO achieves its “robustness.”
WC16
16-Franklin 6, Marriott
Game Theory IV
Contributed Session
Chair: Anastasia Nikolaeva, PhD Student, University at Buffalo,
342 Bell Hall, Amherst, NY, 14260, United States of America,
1 - Bandwagon Investment Equilibrium of a Preemption Game
Kihyung Kim, Postdoc Research Associate, Purdue University,
315 N. Grant Street, West Lafayette, In, 47907, United States of
America, , Abhijit Deshmukh
Given a first mover’s advantage in a competitive market, theoretical research
supports the preemption strategy that invests earlier than competitors. However,
empirical research shows that the preemption strategy is seldom successful, and
simultaneous investments are frequently observed. This research addresses this
gap by deriving the bandwagon investment equilibrium that elucidates how a
firm’s investment entices the competitor’s investment.
2 - Contingency Inventory Reservation Across
Independent Distributors
Sercan Demir, University of Miami, 1251 Memorial Drive,
Department of Industrial Engineering, Coral Gables, FL, 33146,
United States of America, , Murat Erkoc
We study contingency inventory reservation contracts across a single supplier and
multiple buyers. Buyers are distributors operating in independent markets and
are subject to adverse events. They enter into contractual agreement with the
supplier for carrying contingency inventory. The supplier sets the reservation
terms to which the buyers respond with their reservation decisions. We
investigate equilibrium reservation policies under varying probabilities and
market impacts of disruptions.
3 - Vendor Competition in Remanufacturing
Debabrata Ghosh, Assistant Professor, IIM Calcutta, NAB, K-307,
D.H. Road, Joka, Kolkata, WB, 700104, India,
, Sirish Gouda
In this paper we study a case where vendors compete during re-manufacturing
for an OEM. While studies have highlighted competition between retailers under
product take-back, in several emerging economies third party vendors play an
interesting role in re-manufacturing. Using game theoretic set up we study the
decisions of the OEM and vendors under re-manufacturing. Our study reveals
interesting pricing and re-manufacturing decisions under competition.
4 - Selling Format and Contract Type in Electronic Marketplaces with
Product and Retail Competition
Baixun Li, Guangdong University of Finance and Economics,
21 Chisha Road, Guangzhou, Guangzhou, 510320, China,
This paper investigates the choices of selling format and contract type of two
supply chains, which compete at both manufacturer and retailer levels. We
consider two kinds of selling format: agency selling format and reselling format,
and two kinds of contract: wholesale price contract and revenue sharing contract.
We show that the equilibrium selling format of the online retailers depend on the
contract type, channel power, and competitive intensity at both manufacturer and
levels.
5 - Approximating Nash Equilibria in a Two Product
Two Firm Oligopoly
Anastasia Nikolaeva, PhD Student, University at Buffalo, 3
42 Bell Hall, Amherst, NY, 14260, United States of America,
, Mark Karwan
We consider an oligopoly industry and the Cournot-Nash model with two firms
competing on the basis of quantity. In our application there are two
homogeneous products of interest and customers sign a one or joint product
supply contract. Existing customers and a set of new customers not currently
served are competed for by each firm with a goal of profit maximization. We
approximate Nash Equilibria using aspects of Game Theory, Simulation and
Integer Programming and present empirical results.
WC16