2016 INFORMS Annual Meeting Program

TB70

INFORMS Nashville – 2016

TB68 Mockingbird 4- Omni

4 - Product Bundling For Airline Customers Manini Madireddy, Senior Operations Research, Sabre, 3150 Sabre Dr, Southlake, TX, 76092, United States, Manini.Madireddy@sabre.com, Goda Doreswamy, Meisam Hejazi Nia, Ramasubramanian Sundararajan We consider the problem of product bundling (seats and ancillaries) in the context of offering the right products to the right customer at the right price and time, in such a way as to satisfy customer needs and maximize airline revenue. This problem falls in the realm of airline revenue management and retail e- commerce. We present a solution approach to construct, optimize and personalize offers to customers. We demonstrate the utility of our approach through illustrative results on real and simulated data. TB70 Acoustic- Omni Transportation, Planning I Contributed Session Chair: Liang Wang, Phd Candidate, Harbin Institute of Technology, Harbin, China, 14b910008@hit.edu.cn 1 - A Segmented Logistic Regression Model To Construct A Valid Set Of Itineraries From A List Of Weekly Flight Legs Anand Seshadri, Principal, Operations Research, SABRE Airline Solutions, 3rd Floor, Navigator Building, International Tech Park, Bangalore, 560045, India, ug97044@yahoo.com, Gautam Pradhan In this paper, we present a robust approach to rank and remove invalid itineraries and retain a set of good itineraries. Traditionally, this has been accomplished by a heuristic model. The main disadvantage of a heuristic model is that the rules are based on the past behavior of the system and is not dynamic enough to account for changes in airline service variables (alliances, flight departure times etc.). A heuristic model also does not allow the modeler to eliminate itineraries during the building stage leading to inefficient memory utilization. We calibrate the logistic regression model based on a week of historical and schedule data for a major US carrier and compare the results to a rule based model. 2 - Location Of Stations In A One-way Electric Car Sharing System Hatice Çalık, Université Libre de Bruxelles, Université Libre de Bruxelles, Boulevard du Thriomphe, Brussels, 1050, Belgium, hatice.calik@ulb.ac.be, Bernard Fortz We focus on an electric car sharing system where we have a set of customers, each of which wishes to travel from a point of origin to a point of destination at a certain time of the day. The customers can pick up a car from a station close to their point of origin and leave it to a station close to their destination. The location of the stations, the customers to be served, and the stations they will visit need to be decided in a way that maximizes the profit. We provide exact and heuristic methods to solve the problem and conduct computational experiments on newly generated problem instances. 3 - Spatial-temporal Crash Severity Modeling For Aging-involved Crashes: A Case Of Interstate 95 In Florida Aschkan Omidvar, University of Florida, Gainesville, FL, 32611, United States, aschkan@ufl.edu, Arda Vanli, Eren Erman Ozguven This research aims to develop a binary logistic regression model to discover the significant factors affecting the severe crash occurrences for aging drivers. Crash data from two major metropolitan areas, Miami and Jacksonville, for three consecutive years (2010-2012) are extracted, processed and analyzed using Geographical Information Systems (GIS). These data sets are used to determine factors influencing the severity of crashes and compare them with those for other age groups. Next, we investigate the spatial and temporal variation of the effect of the influential variables, on severity of aging-involved crashes by applying variable selection on the fitted logistic regression models.

Tutorial: Wind Energy Applications Sponsored: Quality, Statistics and Reliability Sponsored Session

Chair: Yu Ding, Mike and Sugar Barnes Professo, Texas A&M University, MS 3131, ETB 4016, College Station, TX, 77843, United States, yuding@tamu.edu Co-Chair: Eunshin Byon, University of Michigan, 1205 Beal Avenue, College Station, MI, 48109, United States, ebyon@umich.edu 1 - Tutorial For Wind Energy Data Analytics Yu Ding, Texas A&M University, College Station, TX, 77843, United States, yuding@tamu.edu, Eunshin Byon This tutorial session discusses data analytics issues relevant to wind energy applications. It entails three parts: ): (1) general background of wind energy and data availability; (2) power curve modeling and turbine performance evaluation; and (3) importance sampling and turbine reliability evaluation. The two session chairs will be the co-presenters in this tutorial session. TB69 Old Hickory- Omni Decision Support Systems I Contributed Session Chair: Manini Madireddy, Senior Operations Research, Sabre, 3150 Sabre Dr, Southlake, TX, 76092, United States, Manini.Madireddy@sabre.com 1 - Considering Passenger Recovery In Airline Operations Recovery Jia Kang, Senior Operations Research, Sabre Airline Solutions, The Sabre AirCentre Recovery Manager (Ops) helps airlines quickly recover both the schedule and aircraft rotations from various disruptions (curfews, weather, unplanned maintenances, etc.) by taking into account operational restrictions and several conflicting tradeoffs. In this presentation we introduce a new feature of Recovery Manager called the Passenger Flow Module (PFM) that incorporates passenger re-accommodation decisions during schedule recovery. The generated solutions significantly reduce the impact to passenger flows in airline network as well as overall passenger inconvenience. 2 - Fuzzy Association Rule Mining Framework For Product Selection In E-Commerce Shekhar Shukla, Doctoral Candidate, Indian Institute of Management Lucknow, FPM H-2 Room No. 41, IIM Campus Prabandh Nagar Off Sitapur Road, Lucknow, 226013, India, shekhar.shukla@iiml.ac.in, Ashwani Kumar We present a robust and a unique framework of product selection that incorporates social influence factors and provides a numerical strength of suitability of each available product based on a customer’s set of requirements. The Framework generates Fuzzy Association Rules based on product attributes incorporated with customer reviews as objective weights to these attributes (using Shannon’s Entropy) and market popularity parameters as rule implications. These rules are used as a descriptive model for each product to perform an Association Based Classification. 3 - A Modeling Framework For The Strategic Design Of Local Fresh Food Systems 3150 Sabre Drive, Southlake, TX, 76092, United States, jia.kang@sabre.com, Dinakar Gade, Sureshan Karichery In this work we demonstrate that certain geographical regions might have the potential to produce high-value fresh fruits and vegetables that can be both profitable for current farmers and can incentivize new entrants into local food systems. Specifically, we develop an optimization-based framework that uncovers hidden production capabilities within a region by (1) identifying needed technologies and resources, (2) considering complementary environmental characteristics and market price behavior, and (3) addressing logistic and supply chain planning decisions. It also sets the framework for incorporating parameter randomness. This work addresses research related to local food systems. Hector Flores, PhD Candidate, Arizona State University, 513 W. 17th Street, Tempe, AZ, 85281, United States, hector.flores@asu.edu, Rene Villalobos

287