2016 INFORMS Annual Meeting Program

WE26

INFORMS Nashville – 2016

WE26 110B-MCC Information Systems IV Contributed Session 1 - Information System that Implements Shapley Algorithm For The Evaluation Of The Competitive Value at a Cluster And Enterprise Level Miguel Jimenez, Information systems and technologies leader, Universidad de la Costa, Barranquilla, 08002, Colombia, mjimenez@fcimec.org, Luis Eduardo Ramirez, Stefanie Cortina, Dionicio Neira, William Manjarres De Avila Based on the strategic and financial valuation of enterprises that wish to participate in a cluster, we developed an information system that allows the enterprises evaluate their competiive value with or without their participation in the cluster, by implementing Shapley algorithm to solve a cooperative game model for supply chains. 2 - Advertising Competition With Third Party Cookies Arslan Aziz, Carnegie Mellon University, 5624 Fifth Ave, Apt C16, Pittsburgh, PA, 15232, United States, arslan.aziz7@gmail.com, Rahul Telang Tracking of online consumer behavior by third party data vendors has become ubiquitous. Data from such tracking is made available to advertisers to help increase the returns from targeting. However, such information may also increase competition among advertisers seeking to target the same consumers. We describe a duopoly model of brand advertisers competing for consumers with uncorrelated brand preferences in a second-price auction. We find the conditions under which availability of third party tracked data might reduce brand surplus by intensifying competition. WE27 201A-MCC DMA Business Analytics Contributed Session Chair: Byeong-Yun Chang, Ajou University, San5 Woncheon-dong, Yeongtong-gu, Suwon, 443-749, Korea, Republic of, bychang@ajou.ac.kr 1 - Relationship Between R2 And F Statistic In Linear Regression Nizar Zaarour, Assistant Teaching Professor, Northeastern University, Boston, MA, 02115, United States, n.zaarour@neu.edu, Emanuel Melachrinoudis There are several misconceptions in interpreting the value of R2 in Regression Analysis. R2 is heavily dependent on the sample size n while outliers may skew its value. In this paper, we comment on these observations and express the relationship between the R2 and the F statistic to derive the range of values of R2 that provide consistent results with the Hypothesis Testing of the slope. This analysis is done by considering the Simple Linear Regression case, where there is only one independent variable k. 2 - Heuristic Search For Good Decisions In Generalized Quadratic Assignment Problems Steven Orla Kimbrough, Professor, University of Pennsylvania, 103 Bentley Avenue, Bala Cynwyd, PA, 19004, United States, kimbrough@wharton.upenn.edu, Monique Guignard-Spielberg, Frederic H Murphy We discuss decision sweeping of optimization models, in which we collect a number of judiciously chosen decisions (feasible and infeasible settings of the decision variables) from the larger space of decisions. We focus on the resulting insights, especially with regard to Generalized Quadratic Assignment Problems with soft constraints. In particular, we explore alternative heuristics for generating decisions of interest. 3 - Predicting Urban Blight Using A Data Science Approach Naveen Kumar, University of Memphis, Memphis, TN, United States, nkumar7@memphis.edu, William J Kettinger, Chen Zhang The existence of blighted neighborhoods is detrimental to public health, safety, and economic growth of urban areas. Identifying properties where early blight interventions would result in improvement of neighborhoods can have tremendous value to property owners, policymakers, and society. However, understanding urban blight is a complex problem demanding advanced data science. A wide variety of evolving social, economic, and political factors interact with each other causing the problem. We propose to predict blight incidences using data analytics and recommend early interventions to reduce blight incidences in the City of Memphis. Luis Eduardo Ramirez, Lauren Castro, Lauren Castro, Diana Gineth Ramirez-Rios, Orlando Bustamante,

4 - Does Information Transparency Help Retain Customers? Evidence From The Insurance Industry Zhi Cheng, Temple University, Philadelphia, PA, United States, aaronzhi.cheng@gmail.com, Ting Li, Paul Pavlou This work investigates whether and how information transparency affects customer churn. Two competing theories predict this effect, price elasticity (that induces churn) and high product informedness (that reduces churn). To address this tension, we use a unique dataset from a major European insurance company to show that customers acquired from channels with higher information transparency (a price comparison website) are less likely to churn than those from lower transparent channels by 3%, implying that information transparency helps reduce customer churn. Our findings suggest managers better allocate resources across channels to reduce churn by considering information transparency. 5 - A Study On Stock Prices Forecasting Byeong-Yun Chang, Associate Professor, Ajou University, San 5 Woncheon-dong, Yeongtong-gu, Suwon, 443-749, Korea, Republic of, bychang@ajou.ac.kr, Yucong Chen Stock price forecasting is a very popular issues in nowadays, which can make a contribution to the investing technologies and facilitate those who want to get a better understanding of stocks’ trend line in the future so as to make their investment decisions. this research is going to use a modified two stage EWMA and classical method, ARIMA and MAR to do prediction for companies’ stock price and electricity. For further work, we are going to propose a hybird model which combine TS-EWMA, ARIMA, and MARS. WE28 201B-MCC Retail Operations I Sponsored: Manufacturing & Service Oper Mgmt Sponsored Session Chair: Jan C Fransoo, Eindhoven University of Technology, Eindhoven, Netherlands, j.c.fransoo@tue.nl 1 - Optimal Channel Choices Of Traditional Retail Jiwen Ge, PhD Candidate, Eindhoven University of Technology, Eindhoven, Netherlands, j.ge@tue.nl, Dorothee Honhon, Jan C Fransoo, Lei Zhao Nanostores are small retail stores which are prevalent in the mega-cities of emerging markets. We consider one CPG manufacturer selling one product to a cluster of nanostores either via the wholesale or the direct channel. We provide optimality conditions for each channel strategy when market demand is constant or grows deterministically within a finite time horizon. 2 - Coordinated Delivery To Nanostores In Megacities Ruidian Song, Tsinghua University, Beijing, 100084, China, srd13@mails.tsinghua.edu.cn, Lei Zhao, Jan C Fransoo, Tom Van Woensel In megacities in emerging economies, there exists a large amount of independently operated, traditional format, small grocery stores (nanostores). The limitation in store space and cash flow force them to order frequently with small order sizes, which results in high delivery cost. We study potential strategies to coordinate these deliveries and examine their impact. 3 - Demand Estimation Under Multi-store Multi-product Substitution In High Density Traditional Retail Tianhu Deng, Tsinghua University, Beijing, China, deng13@tsinghua.edu.cn, Mingchao Wan, Lei Zhao, Jan C Fransoo In large cities in emerging economies, traditional retail is present in a very high density, with multiple independently owned small stores in each city block. Consequently, when faced with a stockout, consumers may not only substitute with a different product in the same store, but also switch to a neighboring store. We study this problem using both Nested Logit Model and Exogenous Model. Furthermore, we estimate the parameters of the two models using a Markov chain Monte Carlo algorithm in a Bayesian manner. We numerically find that the Nested Logit model outperforms the Exogenous Substitution model in estimating substitution probabilities.

492