Informs Annual Meeting Phoenix 2018

INFORMS Phoenix – 2018

MB21

2 - Price Optimization under an N-pack Choice Model Guang Li, Queen’s University, Smith School of Business, Kingston, ON, Canada, Ying Cao We consider the price optimization problem for a retailer under an n-pack choice model. Under such a model, each customer would either purchase n items from a collection of m products or leave with no purchase. Given an offered assortment, the retailer aims at maximizing his total revenue by setting the right price for each product. We study the structural properties of the optimal prices under different pricing schemes and develop efficient algorithms for price optimization 3 - Operations Management under Sequential Choice Models Ruxian Wang, Johns Hopkins University, Carey Business School, 100 International Dr, Baltimore, MD, 21202, United States Consumers may follow a sequence to choose the products they would like to choose. We study the operations management problems and derive useful managerial insights. 4 - Assortment Planning with N-pack Purchasing Consumers Ying Cao, University of Texas at Dallas, Dallas, TX, United States, Dorothee Honhon For many product categories, customers often buy multiple differentiated products on a given store visit for staggered consumption until the next store visit. Such customers are referred to as n-pack purchasing customers in Fox et al (2017). We consider a retailer who makes product assortment decisions in a given product category facing n-pack purchasing customers. We study the structural properties of the optimal assortment under two different customer choice rules. And we explore how the retailer’s assortment decision and total profits are impacted when the retailer ignores the “choice premium which captures the utility that consumers derive from variety in their shopping basket. n MB20 North Bldg 129A Joint Session RMP/Practice Curated: Sharing Economics Sponsored: Revenue Management & Pricing Sponsored Session Chair: Han Zhu 1 - Courteous or Crude? Operational Tools to Shape User Behavior in Ride-sharing Platform with Bilateral Ratings Yunke Mai, Duke University, Durham, NC, United States, Yuhao Hu, Zilong Zou, Bin Hu, Sasa Pekec We study the impact of riders’ manners and drivers’ selectiveness on the performance of ride-sharing platforms. We develop an evolutionary game-theory model to study how user behavior on the platform evolve over time under a bilateral rating system. We identify evolutionary trajectories of user behavior and stable equilibria for the platform usage as a function of the system parameters. In addition, we investigate how platform’s operational decisions such as pricing, supply/demand management, and choice of matching protocols impact user behavior and platform’s long-run performance. 2 - Analysis of Incentivized Ride Matching as Stackelberg Queue Shumin Ma, The Chinese University of Hong Kong, ERB615, CUHK, Hong Kong, Qi Wu We study incentive strategies using a queueing game approach. Our key assumption is that the driver supply is finite and reusable while riders arrive stochastically. With this assumption, we first establish the endogenous forces driving the imbalances between supply and demand with zero monetary incentive. We then administer incentives between the platform and the driver population via Stackelberg games and study the system’s intrinsic capacity bounds in steady state. We show that the optimal amount of myopic incentives is achieved when the circulation of the reusable pool of the driver supply is the fastest. Further spending beyond the optimal, however, is potentially disruptive. 3 - Dynamic Type Matching Yun Zhou, McMaster University, 1280 Main Street West, Hamilton, ON, L8S 4M4, Canada, Ming Hu Motivated by the sharing economy, we consider a dynamic matching problem with heterogeneous supply and demand types. This paper studies the optimality and near-optimality of matching policies under a given priority rule. For two cases with vertical and horizontal types, respectively, we characterize the optimal prioritized matching policy.

4 - Short-term Asset Rentals and Corporatization of Platform Pricing Han Zhu, McGill University, Montreal, QC, H3H 1K4, Canada, Mehmet Gumus, Saibal Ray Recently, we have seen the emergence of short-term rentals platforms. In this paper, we focus on how the platform decides on the price to charge their customers. Specifically, in a platform like AirBnB, the price is effectively set based on a market mechanism that matches supply and demand. But, some other platforms are more active. They take turnkey control of the assets and determine the price on behalf of the owners that maximizes their profits. Our primary goal in this paper is to understand the implications of this difference in pricing strategy for the direct stakeholders of the platform such as customers, owners and the platform as well as indirect stakeholders such as long-term rentals and hotels. n MB21 North Bldg 129B Advances in Demand Learning for Revenue Management Sponsored: Revenue Management & Pricing Sponsored Session Chair: Pavithra Harsha, IBM Research, Yorktown Heights, NY, 10598, United States 1 - Choice Model Trees: A Joint Framework for Market Segmentation and Decision Making Ryan McNellis, Columbia University, 540 West 122nd Street, Apartment 6C, New York, NY, 10027, United States, Mohammed Ali Aouad, Prasad Chalasani, Adam Elmachtoub, Kris Johnson Ferreira, Michael R. Young We propose a new method for incorporating feature information into marketing decision-making problems. Relevant applications include the recommendation of personalized product assortments, personalized pricing, and customizing bids for advertising exchanges. Our method uses a decision tree to segment the market (e.g., customers), and a choice model involving the decision variable (e.g., product assortment) is then fit locally in each segment. The resulting model is interpretable and easily visualized. We propose a new training algorithm which directly optimizes the likelihood of the resulting collection of choice models. Modifications are explored for improved scalability. 2 - A Model-based Embedding Technique for Segmenting Customers Ashwin Venkataraman, Harvard University, Cambridge, MA, United States, Srikanth Jagabathula, Lakshminarayanan Subramanian We consider the problem of segmenting a large population of customers into non- overlapping groups with similar preferences, using diverse signals such as purchases, ratings, clicks, etc. over a large universe of items, when each customer provides only a few signals. We propose a model-based embedding technique which takes the customer observations and a probabilistic model class generating the observations as inputs, and outputs an embedding—-a low-dimensional representation in Euclidean space—-for each customer. We then cluster the embeddings to obtain the segments. We demonstrate the speed and performance of our method in two case studies including a real implementation on eBay data. 3 - Dynamic Pricing of Limited Inventories with Product Returns Xing Hu, University of Oregon, 1208 University Of Oregon, 484 Lillis, Eugene, OR, 97403, United States, Zhixi Wan, Nagesh N. Murthy Many online retail channels face high rates of product returns. This poses a new challenge to the sellers’ dynamic pricing problem when some returns in good condition can be resold in the selling season. To study the impact of product returns and guide sellers to adjust pricing policies, we build a product returns model by augmenting the classic monopolist’s dynamic pricing framework. We address the technical challenges both analytically and numerically. Our analysis finds that ignoring returns leads to over-pricing and can cause significant revenue loss. The analysis yields easy-to-implement heuristic policies that have good and robust performance relative to the theoretical benchmarks.

155