Informs Annual Meeting 2017

SB22

INFORMS Houston – 2017

2 - Computer-Aided Diagnostic Models for Breast Cancer Screening Oguzhan Alagoz, University of Wisconsin-Madison, 3242 Mechanical Engineering Building, 1513 University Avenue, Madison, WI, 53706, United States, alagoz@engr.wisc.edu Mammography is used for early breast cancer diagnosis however similarities between early signs of breast cancer and normal structures in images make diagnosis a difficult task. We have developed computer-aided diagnostic models and tested them using real-life mammography data. In this presentation, we describe our experiences, list the limitations of the existing CADx models, and provide possible future research directions. 3 - Constraint Relaxation Extensions for Imbalanced Classification Talayeh Razzaghi, Assistant Professor, New Mexico State University, MSC 4230, P.O. Box 30001, Department of Industrial In this work, we study several algorithmic modifications of relaxed support vector machines to tackle a number of disease diagnosis problems with imbalanced datasets in the presence of outliers. We provide computational evidence proving better performance for the combined use of cost sensitive learning with constraint relaxation compared to bagging-based algorithms, which treat imbalanced-ness and outliers separately. 4 - Current Trends and Challenges of Computer-Aided Medical Diagnosis Juri Yanase, Complete Decisions, LLC, 10517 Springbrook Ave., Baton Rouge, LA, 70810, United States, jurijuriy@aol.com, Evangelos Triantaphyllou Computer-aided diagnosis is already more than 50 years old and has entered a maturity stage. It offers certain advantages for employing information from big data and being consistent. It finds applications in a wide range of areas. However, some serious challenges persist and include the need to analyze ultra large volumes of data, and algorithmic, security, and legal liability issues. Engineering, Las Cruces, NM, 88003-8001, United States, talayehr@nmsu.edu, Onur Seref, Petros Xanthopoulos Chair: Georgia Perakis, Massachusetts Institute of Technology, Cambridge, MA, 02142-1347, United States, georgiap@mit.edu 1 - Optimizing Over Tree Based Ensemble Objective Functions Max R. Biggs, Massachusetts Institute of Technology, 77 Massachusetts Ave NE49-40497, Apartment 2, Cambridge, MA, 02139, United States, maxbiggs@mit.edu In many real world applications, there is an unknown relationship between decisions and outcomes. For example, we might have data on previous actions taken in different situations, and observe outcomes. Using this data, there are powerful machine learning algorithms which can be used to build a predictive model. However, there has been little attention on how to make decisions if we fit a tree based ensemble model and use it as our objective function. We will show that this optimization problem can be modeled as a MILP, and present tractable decompositions and heuristics. We apply this methodology to a jury selection problem, where there is uncertainty as to how a juror will influence the jury. 2 - Stochastic Probing to Maximize the Value of Diagnostic Information Mathieu Dahan, MIT, Cambridge, MA, United States, mdahan@mit.edu, Saurabh Amin, Georgia Perakis We consider an adaptive probing problem to maximize the detection of failures in infrastructure networks. We study a stochastic programming model which accounts for the spatial distribution of alerts, including relative distances between locations and predictive accuracy of failure events within each location. Our result shows how to optimally sequence the probing of locations based on the tradeoff in exploiting a given location versus exploring other locations. We connect our work with recent literature on stochastic orienteering. Finally, we apply our approach to improve the value of diagnostic information for a regulated Daniel Chen, Massachusetts Institute of Technology, Cambridge, MA, United States, dcchen@mit.edu, Retsef Levi, Georgia Perakis We describe theoretical and data-driven work in collaboration with an online retailer. In a warehouse, multi-item orders take up capacity on a wall until all items are picked. To avoid exceeding capacity, the retailer must balance picking efficiency with completing orders quickly. Motivated by this, we model an NP- gas utility who invests in leak surveys to detect network failures. 3 - Submodular Batch Scheduling for Warehouse Picking SB22 342D New Problems in Revenue Management Sponsored: Revenue Management & Pricing Sponsored Session

hard batch scheduling problem to minimize average order completion times, where the processing time of each batch is given by a submodular function, generalizing previous problems in the literature. We provide integer programming formulations and approximation algorithms, and testing on a data-driven simulation suggests a potential 15% reduction in the time orders stay at the wall.

SB23

342E New Models in Pricing and Revenue Management Sponsored: Revenue Management & Pricing Sponsored Session Chair: Sandun Perera, PhD, University of Michigan-Flint, 2200 Riverfront Center, 303 East Kearsley Street, Flint, MI, 48502, United States, sperera@umich.edu 1 - Multi-product Pricing with Stockouts and Satisficing Customers Varun Gupta, Penn State Erie, The Behrend College, 5101 Jordan Rd, Burke 281, Erie, PA, 16563, United States, vxg15@psu.edu, Metin Cakanyildirim Stockouts for high inventory turnover products lead to loss of sales as customers may substitute their preferred product (stocked out) with another product (available). We study centralized pricing for a retailer and equilibrium prices for competing retailers selling to satisficing customers with stockout-based substitution under lost sales and backorders. 2 - Dynamic Pricing and Replenishment for Seasonal and Regular Products Oben Ceryan, LeBow College of Business, Drexel University, 2040 Market St. Apt. 801, Philadelphia, PA, 19103, United States, oceryan@drexel.edu We consider a firm that offers two substitutable products that differ in how their inventories are managed, a seasonal product with a fixed initial quantity that allows dynamic price adjustments but no replenishments, and a regular product with a static price that can be periodically replenished. We study the impact of these asymmetries on optimal dynamic pricing and replenishment decisions. 3 - Dynamic Pricing to Explore Markets with Customer- and Time- heterogeneity Meng Li, Rutgers University, 227 Penn Street, Camden, NJ, 08102, United States, meng.li@rutgers.edu, Bora Keskin Consumers are often heterogeneous in their preferences for product quality, and companies usually face uncertainty about consumer preferences when they sell differentiated products to such heterogeneous consumers. We study this problem Sandun Perera, School of Management, The University of Michigan-Flint, 2200 Riverfront Center, 303 E. Kearsley Street, Flint, MI, 48502-1950, United States, sperera@umich.edu, Syagnik Banerjee, Metin Cakanyildirim At times of crisis and disaster, donors often pay for necessary commodities distributed via local retailers. Retailers need to set prices that balance the tradeoffs between vulnerable customer needs and their ability to pay. We study this pricing problem under different market conditions. 342F Pricing and Consumer Experience in Behavioral and Service Settings Sponsored: Revenue Management & Pricing Sponsored Session Chair: Sami Najafi-Asadolahi, Santa Clara University, Santa Clara, CA, 95053, United States, snajafi@scu.edu 1 - Why Markdown as a Pricing Modality? the Role of Threat of Entry Elodie Adida, University of California Riverside, Riverside, CA, United States, elodieg@ucr.edu, Ozalp Ozer Markdown as a pricing modality is ubiquitous in retail. Despite its practical advantages everyday-low-price is rarer. We explore whether and why either of these pricing modalities is an effective defense against a firm that enters the market with the alternative pricing modality when consumers are heterogeneous, strategic and regret-prone. in a situation where a company dynamically optimizes its prices. 4 - Optimizing Starvation to Avoid Cannibalization: Retailer’s Strategies for Markets in Crisis SB24

42