Informs Annual Meeting 2017

SD53

INFORMS Houston – 2017

3 - The Centre Cannot Hold: Fragmentation in Multisided Markets Abhishek Roy, University of Texas-Austin, 2110 Speedway Stop B6500, Austin, TX, 78712-1277, United States, abhishek.roy@utexas.edu, Edward G. Anderson, Geoffrey Parker Existing literature on multi-sided markets, or platforms, examine the effect of opening up the platform to third-party developers on the platform sponsor. However, in some cases, opening up of certain aspects of a platform results in a fragmented platform that causes negative externalities on other third-party developers or sellers. We develop a model to analyze the effect of such fragmentation in the context of mobile operating systems, where the device side of the ecosystem is opened up, on the application developers, and in turn, on the platform itself. 4 - Ad-Blockers, Advertisers & Internet - Economic Implications of Ad Blocking Abhishek Ray, Purdue University, 401S, Grant Street, West Lafayette, IN, 47907, United States, ray52@purdue.edu, Hossein Ghasemkhani, Karthik Kannan The rise of ad-blockers as two-sided platforms has prompted discussions on whether ad-blockers are good for the online ecosystem. Considering the structure of interactions among the end-users, ad-blockers charge advertisers membership fees for certifying quality of their ads (whitelisting) and letting them through, but are free for the users. Our main research question is to study this mechanism and investigate whether it is optimal. Using a game-theoretic model using which we provide answers to questions such as which side (user or advertiser) should be charged by the ad-blocker, the impact of optimal price structure on end-user participation and overall implications for ad quality online. 361E Recent Innovations in Last-Mile Urban Deliveries Sponsored: TSL, Urban Transportation Sponsored Session Chair: Bo Zou, University of Illinois, Chicago, IL, 60607, United States, bzou@uic.edu 1 - Designing Mechanisms for Efficient Crowdsourced Urban Parcel Delivery SD52 This paper presents a mechanism design based approach to crowdshipping. A Delivery Service Provider solicits private information from individuals such as willingness-to-ship and available time window, based on which to assign shipments with minimum total shipping cost. The mechanisms recognize that individuals can misreport private information to gain self-interest, and devise joint shipment assignment-payment schemes to incentivize truthful participation of individuals. Both static and dynamic cases are investigated. Numerical results with a case study of the near north side of Chicago demonstrate the cost advantage of the mechanisms over the state-of-the-practice crowdshipping. 2 - Crowdshipping Consolidation in Urban Logistics Jane Lin, University of Illinois-Chicago, 842 W. Taylor Street (M/C 246), Chicago, IL, 60607, United States, janelin@uic.edu, Sudheer Ballare Crowdshipping involves making use of everyday individuals with spare time and capacity to fulfill the variable delivery demand generated by e-commerce. This study performs preliminary investigation of an existing crowdsourcing delivery company with respect to the operational factors such as package size, delivery distance, demand frequency and distribution, the user characteristics including customer and driver profiles, and the pricing model. We further investigate the opportunities of consolidating crowdshipping tasks to further reduce cost and environmental impact. The findings will shed light on the market demand trending and growth opportunities in crowdsourcing deliveries. 3 - Reliability and the Edge of Vulnerability in Crowdsourced Shipping Alireza Ermagun, Post-doc, Northwestern University, 2145 Sheridan Rd, Evanston, IL, United States, Alireza.ermagun@northwestern.edu, Amanda Stathopoulos This study introduces reliability indices to measure the vulnerability of crowd- shipping as an innovative urban delivery system. This is a critical matter in the start-up phase of crowd-shipping services, as failing in providing a reliable system discourages registered customers and dissuades other potential customers to use the service. The results are informative to understand the caveats of the system in detail and to implement effective policies to augment the reliability where it is needed and leverage the resources across the system. Bo Zou, University of Illinois at Chicago, 2073 Engineering Research Facility, 842 W. Taylor Street, Chicago, IL, 60607, United States, bzou@uic.edu, Nabin Kafle

4 - Multi-channel Network Planning Vincent Karels, Eindhoven University of Technology, Eindhoven, Netherlands, V.C.G.Karels@tue.nl

Investigations are made on whether different channels (e-tail, detail and retail) can be better integrated in planning, scheduling and execution. Resources can be effectively shared over and within the three channels. A framework facilitating this integration, with an emphasis on routing, is introduced. This framework is subsequently applied on test- and real-life instances. From these instances it can be shown that integration leads to large benefits. 5 - Bidding, Accepting, and Delivering: On Performance Assessment of Crowd-shipping Alireza Ermagun, Northwestern University, Evanston, IL, United States, alireza.ermagun@northwestern.edu, Amanda Stathopoulas This study assesses the performance of crowd-shipping as an innovative urban delivery system. We develop a set of hazard-based duration models to explore socioeconomic, built-environment, and intra-system characteristics that describe the bidding, accepting, and delivering status of packages as a function of time. The results from a real crowdshipping market operating over two years are informative to design policies and incentives relating to prediction, pricing, or promotion for delivery performance, and thereby to improve the efficiency and acceptability of crowd-shipping systems. 361F Emerging Data Sources and Travel Demand Modeling II Invited: TSL, Intelligent Transportation Systems (ITS) Invited Session Chair: Ali Arian, University of Arizona, 415 N Park Avenue, Tucson, AZ, 85719, United States, arian@email.arizona.edu 1 - Real Time Transportation Mode Detection with Wearable Devices and Neural Networks Young-Ji Byon, Associate Professor, Khalifa University of Science Technology, 216F Al Saada St and Muroor Rd, P.O. Box 127788, Abu Dhabi, 127788, United Arab Emirates, youngji.byon@kustar.ac.ae, Chung-Suk Cho, Jun Su Ha, Heungjo An, Taeyeon Kim For traffic monitoring purposes, Departments of Transportation collects data from various dedicated sensors and probe vehicles throughout the road networks. If the modes of transportation the users of smartphones are in can be determined, DOTs can collect mode-specific data from the users as if they are probes themselves. In general, a smartphone is embedded with an accelerometer and a GPS sensor. Emerging wearable devices can provide multiple streams of acceleration values that can enhance mode-detection performances. It is found that having access to acceleration values of multiple parts of the body of a traveler helps identifying the modes 2 - Community Detection with Similar and Dissimilar Travel Activity Patterns Ali Arian, University of Arizona, 415 N. Park Avenue, Tucson, AZ, 85719, United States, arian@email.arizona.edu, Alireza Ermagun, Yi-Chang Chiu This research introduces a model to detect communities benefiting from the similarities and dissimilarities in the activity patterns of individuals based on an activity-network. We test a set of community detection methods trip information of 438 individuals using an incentive-based mobility app, Metropia, over 300 days in Tucson, Arizona. The model is validated in predicting the carpooling behavior of users. 3 - Travel Behavior Classification with Social Networkand Deep Learning Yu Cui, University at Buffalo, 204c Ketter Hall, 429-7333, SUNY Buffalo, NY, 14228, United States, ycui4@buffalo.edu, Qing He, Alireza Khani, Qian Wang We build a social network for all travelers and employ community detection algorithm to cluster travelers based on the identified similarities of travel patterns. We further create an image of activity map from each traveler’s activity matrix. Finally, a deep learning approach with convolutional neural network is employed to classify travelers into corresponding groups according to their activity maps. 4 - Passenger Behavior Analysis in Beijing Metro Network using AFC Data and Agent-based Simulation Alireza Khani, University of Minnesota, Minneapolis, MN, United States, na, Kai Lu, Baoming Han AFC data in Beijing metro system has both entry and exit time and location stamps for each passenger. This study based on Beijing AFC data, calibrates the parameters for trip chaining model with the statistical and cluster methods and finds out the passenger trip chain pattern in Beijing network. SD53

119