2015 Informs Annual Meeting

MA44

INFORMS Philadelphia – 2015

MA44 44-Room 103B, CC Data Driven Pricing Sponsor: Revenue Management and Pricing Sponsored Session

better profits than others. In this talk, we will discuss some practical issues frequently encountered in the modeling of price and revenue optimization, and we will go through several real world examples to illustrate some of the best practices. 2 - Revenue Functions: Demand Aggregation for Fleet Allocation in Car Rental Industry Manu Chaudhary, Scientist, PROS, 3100 Main Street, Suite #900, Houston, TX, 77002, United States of America, mchaudhary@pros.com Fleet allocation for car rental industry is a multi-dimensional, large scale network optimization which is computationally expensive. We address this complexity by introducing an intermediate optimization that generates Revenue Functions which aggregate the demand forecast to a higher level. This reduces the dimensional complexity of the modified fleet allocation problem significantly and makes it feasible to run in a live production system. 3 - Identifying a More Accurate Historical Data Subset from a Noisy Historical Dataset: A Forecasting Example Gregory Vogel, Manager, Revenue Science, Holland America Line, 300 Elliott Ave W, Seattle, WA, 98119, United States of America, gvogel@hollandamerica.com When looking at similar products that have overlapping booking periods, noise is common. It is common practice to compile a dataset for processes such as demand forecasting by compiling the complete set of relevant history. We ask the question, can we identify a subset of history that will produce a more accurate forecast? We utilize a cruise example to demonstrate and improved forecast and present the method developed. 4 - Joint Optimization of Pricing and Marketing for Globally Maximum Profit Sharon Xu, UCLA Statistics Department, 8125 Math Sciences, Los Angeles, CA, 90095, United States of America, sharon.xu@ucla.edu With the recent influx of high-granularity data, businesses are able to achieve a more comprehensive understanding of their customers. To better leverage this data, we present a new way to maximize profit by jointly optimizing pricing and advertising spend. We first create a predictive model to quantify how pricing and advertising influence consumer purchase decisions, then use this to inform a model that simultaneously optimizes pricing and advertising decisions to obtain the maximum profit. 5 - An Airline RM Model with Capacity Sharing Ang Li, Scientist, PROS, Inc., 3100 Main St. #900, Houston, TX, 77002, United States of America, ali@pros.com, Darius Walczak We consider a single-leg airline RM problem with shared seating capacities between business and economy compartments. In particular, a curtain that separates the two compartments can be installed on the day of flight departure. We solve the control problem as a dynamic program that jointly considers the booking levels on both sides of the curtain. We then compare the optimal revenue achieved to an alternative scenario with an optimal curtain position but without capacity sharing. MA46 46-Room 104A, CC Healthcare Operations Sponsor: Manufacturing & Service Oper Mgmt/Service Operations Sponsored Session Chair: Carri Chan, Columbia Business School, 3022 Broadway, Uris Hall, Room 410, New York, NY, 10027, United States of America, cwchan@columbia.edu 1 - Service Decisions with Two-dimensional Customer Heterogeneity Tolga Tezcan, Associate Professor, London Business School, Regent’s Park, London, UK, NW14SA, United Kingdom, ttezcan@london.edu, Balaraman Rajan, Avi Seidmann In this work we analyze the operational decisions of a server dealing with customers who are heterogeneous on two dimensions. We apply the results in the context of a specialist seeing patients suffering from chronic conditions and patients who differ in their preferences to a newly introduced telemedicine technology. Our results help analyze telemedicine adoption and optimal decisions for a service provider.

Chair: Wedad Elmaghraby, Associate Professor, University of Maryland, University of Maryland, 4311 Van Munching Hall, College Park, MD, 20742, United States of America, welmaghr@rhsmith.umd.edu Co-Chair: Shawn Mankad, Assistant Prof Of Business Analytics, University of Maryland, 4316 Van Munching Hall, College Park, MD, 21201, United States of America, smankad@cornell.edu 1 - More than Just Words: on Discovering Themes in Online Reviews to Explain Restaurant Closures Shawn Mankad, Assistant Prof Of Business Analytics, University of Maryland, 4316 Van Munching Hall, College Park, MD, 21201, United States of America, smankad@cornell.edu, Anand Gopal, Jorge Mejia We complement the existing research on online reviews by proposing a novel use of modern text analysis methods to uncover the semantic structure of online reviews and assess their impact on the survival of merchants in the marketplace. We analyze online reviews from 2005 to 2013 for restaurants in a major metropolitan area in the United States and find that variables capturing semantic structure within review text are important predictors of the survival of restaurants. 2 - Cost-per-impression Pricing for Display Advertising Sami Najafi-Asadolahi, Santa Clara University, 500 El Camino Real, Santa Clara, CA, United States of America, snajafi@scu.edu We consider a web publisher posting display ads on its website and charging based on the cost-per-impression pricing scheme. The publisher is faced with uncertain demand for advertising slots and uncertain supply of visits from viewers. Advertisers choose ad campaigns that specify their targeted viewers. We determine the publisher’s optimal price to charge and show that it can increase in the number of impressions, which is in contrast to the quantity-discount commonly offered in practice. 3 - Pricing Personalized Bundles: A New Approach and Industrial Application Zhengliang Xue, IBM, Yorktown Heights, NY, United States of America, IBM, zxue@us.ibm.com, Zizhuo Wang, Markus Ettl We optimize the pricing strategies for personalized bundles where a seller provides a variety of products using which customers can configure a bundle and send a request-for-quote. The seller, after reviewing the RFQ, has to determine a price based on customers’ willingness to buy. Such problems are difficult because of the potential unlimited possible configurations of the bundle and the correlations among individual products. We propose a new approach and show the business value by real data. 4 - Measuring the Effects of Advertising: The Digital Frontier Justin Rao, Researcher, Microsoft Research, 641 Avenue of Americas, New York, NY, 10014, United States of America, Justin.Rao@microsoft.com, Randall Lewis, David Reiley Online advertising offers unprecedented opportunities for measurement. A host of new metrics have become widespread in advertising science. New experimentation platforms open the door for firms and researchers to measure the true causal impact of advertising. We dissect the new metrics, methods and computational advertising techniques currently used by industry researchers, highlighting their strengths and weaknesses, and discuss the novel analyses could impact the advertising market.

MA45 45-Room 103C, CC RM in Practice Sponsor: Revenue Management and Pricing Sponsored Session

Chair: Wei Wang, Scientist, PROS, Inc., 3100 Main Street, Suite #900, Houston, TX, 77002, United States of America, weiwang@pros.com 1 - Modeling Issues and Best Practices in Price and Revenue Optimization Yanqi Xu, Director Of Applied Technology, Princess Cruises, 24305 Town Center Rd, Valencia, CA, 91355, United States of America, yxu@princesscruises.com Price and revenue optimization is proven to be critical to improving the top lines in various industries where relatively fixed capacity has to be used to satisfy fluctuating demand. However, there exist ways to set up the models to achieve

160