Regular Session


Confirmed speakers are below. More speakers are coming soon...


Session 1

Title: Belt and Road: a hinterland logistics perspective

Speaker: Jan C. Fransoo, Kuehne Logistics University

Abstract:
The Belt and Road Initiative connects China to a vast continental hinterland. In this presentation, I will interpret our research findings of extensive modeling and empirical work on European small-scale hinterland logistics, to China's large scale hinterland logistics on the Belt and Road.


Title: Optimal learning for urban delivery fleet allocation

Speaker: Lei Zhao, Tsinghua University

Abstract:
In a two-tiered city logistics system, an urban logistics company usually partitions the urban area into regions and allocates its delivery fleet (e.g., vehicles, couriers) to these regions. On a daily basis, the delivery station in each region receives the delivery packages from the city distribution centers and delivers them to customers within the region, using its allocated delivery vehicles. A tactical decision in such a city logistics system is the allocation of its delivery fleet to the regions to minimize the expected operational cost of the entire system. However, due to the complexity of the urban delivery operations and the day-to-day variance of the customer demand, an accurate evaluation of the expected operational cost associated with an allocation decision can be very expensive. We propose a learning policy that adaptively selects the fleet allocation to learn the underlying expected operational cost function by incorporating the value of information. Specifically, we exploit the monotonicity of the expected operational cost in the number of allocated delivery vehicles in a region and extend the idea of knowledge gradient with discrete priors (KGDP) with resampling and re-generation (KGDP-R&R). Our numerical results demonstrate the effectiveness of KGDP-R&R against other learning policies as well as its managerial implications as compared to heuristics in practice.


Title: The logic of matching in ride sharing markets: revenues, service ratings or pick-up times?

Speaker: Hai Wang, Singapore Management University

Abstract:
We study a class of multi-period multi-objective online optimization problems, where a decision maker takes actions over time in an online fashion without being informed of future scenarios. To balance the trade-offs between different objectives, we develop an efficient online policy to derive the "compromise" solution, which minimizes the l_p-distance from the attained KPIs to the utopia target. We apply the online policy in ride sharing market settings, and observe that all parties in the ride-sharing eco-system, from drivers, passengers, to the platform, are better off under our proposed online matching policy.


Title: Velocity-based storage assignment in semi-automated storage systems

Speaker: Rong Yuan, JD.com

Abstract:
Our research focuses on the storage decision in a semi-automated storage system, where the inventory is stored on mobile storage pods. We characterize the maximum possible improvement from applying a velocity-based storage policy in comparison to the random storage policy. We show that class-based storage with two or three classes can achieve most of the potential benefits and that these benefits increase with greater variation in the pod velocities. Through simulation, we observe an 8% to 10% reduction in the travel distance with a 2-class or 3-class storage policy.



Session 2

Title: Fluctuation scaling in large service systems

Speaker: Xiaowei Zhang, Hong Kong University of Science and Technology

Abstract:
Operational decision making in service systems often depends largely on the characterization of the random fluctuations involved. Exogenous arrivals (e.g., customers, orders, etc.) represent a primary source of uncertainty and their stochastic behavior needs to be modeled carefully. In this talk, we will present a new statistical finding regarding the random fluctuations of the arrival process in large service systems, and propose a tractable model accordingly. When a service system under the new arrival model is scaled up, its dynamics is fundamentally different from that typical queueing analysis stipulates, and leads to a new staffing rule for managing the servers. At last, we will demonstrate via data-driven simulation that our staffing rule improves the system performance substantially in general.


Title: Data-driven storage assignment in mobile fulfillment systems

Speaker: Chen Wang, Tsinghua University

Abstract:
We study the bundling storage strategy in mobile fulfillment systems by extracting the structure of demand from e-commerce orders. Bundling SKUs that are frequently purchased together can substantially lower the number of pod fetches per order. We propose a novel estimation-optimization scheme where the penalized linear-effect model in machine learning is applied to depict customer purchase patterns, and the estimation results are exploited to derive the bundling policy of SKUs. Efficient computational algorithms are developed for the estimation and optimization models to accommodate large-scale, sparse and dynamic e-commerce orders. We find bundling storage is necessary when there is large variability in the co-occurrence rates of different SKU combinations; otherwise the random assignment policy suffices. Even if coupled with the velocity-based pod layout strategy, the proposed bundling policy still generates considerable additional benefits. However, such advantages are less pronounced when we take into account order batching and delays due to replenishment operations.


Title: An analysis of "buy x get one free" customer reward program

Speaker: Yan Liu, Tianjin University

Abstract:
Redemption hurdles are essential yet little understood design components of customer reward programs. We model and analyze two types of redemption hurdles, i.e., redemption threshold and expiry policy, in the context of “buy X, get one free” (BXGO) reward program, where a consumer earns a free product after she accumulates X purchases (redemption threshold) and must redeem before the reward points expire (expiry policy). We further incorporate recent research findings that consumers derive nontrivial psychological utility from a free product (we call it “transaction utility”), in addition to the economic utilities it confers. In our model, consumers with heterogeneous valuations and purchase frequencies interact with a monopolistic seller over the long-run. We analytically show that both types of redemption hurdles create additional incentive for the consumer to purchase. This finding not only rationalizes the “goal gradient hypothesis” documented in the behavioral literature, but also allow us to precisely quantify such positive effects created by redemption hurdles. Building on the insights from the consumer analysis, we then explore the optimal design of redemption hurdles along with the optimal price. Several interesting results emerge. First, despite its simplicity, a BXGO program can significantly increase firm profitability, and potentially create a win-win outcome for the firm and the consumers. Second, a BXGO program can be more profitable than a cash reward program without reward accumulation for a modest level of transaction utility. Third, a finite expiration term is not a necessary condition for the effectiveness of BXGO programs, which offers a possible explanation why many BXGO program rewards do not carry an expiration date.


Title: Pricing and inventory management with operational data analytics

Speaker: Qi Annebelle Feng, Purdue University

Abstract:
We consider the problem of deciding the selling price and order quantity for a newsvendor given data on the past selling price and demand. Particularly, we consider the case where we know the structural model with two parameters, scale and demand elasticity for the demand. We first discuss the classical approach to solving this problem, and then compare it to the operational data analytics approach. Depending on the knowledge of the parameters, and decision variables, we discuss different validation and data integration models for operational data analytics approach and show that they perform better than the classical approach


Organizer


Co-organizer


Copyright © 2018, Tsinghua-JD-POMS-2018.

Please contact us with your questions, comments, and suggestions.