Search

Problem definition: We study an urban bike lane planning problem based on the fine-grained bike trajectory data, which are made available by smart city infrastructure, such as bike-sharing systems. The key decision is where to build bike lanes in the existing road network. Academic/practical relevance: As bike-sharing systems become widespread in the metropolitan areas over the world, bike lanes are being planned and constructed by many municipal governments to promote cycling and protect cyclists. Traditional bike lane planning approaches often rely on surveys and heuristics. We develop a general and novel optimization framework to guide the bike lane planning from bike trajectories. Methodology: We formalize the bike lane planning problem in view of the cyclists’ utility functions and derive an integer optimization model to maximize the utility. To capture cyclists’ route choices, we develop a bilevel program based on the Multinomial Logit model. Results: We derive structural properties about the base model and prove that the Lagrangian dual of the bike lane planning model is polynomial-time solvable. Furthermore, we reformulate the route-choice-based planning model as a mixed-integer linear program using a linear approximation scheme. We develop tractable formulations and efficient algorithms to solve the large-scale optimization problem. Managerial implications: Via a real-world case study with a city government, we demonstrate the efficiency of the proposed algorithms and quantify the trade-off between the coverage of bike trips and continuity of bike lanes. We show how the network topology evolves according to the utility functions and highlight the importance of understanding cyclists’ route choices. The proposed framework drives the data-driven urban-planning scheme in smart city operations management.

This paper presents a column-and-constraint generation algorithm for two-stage stochastic programming problems. A distinctive feature of the algorithm is that it does not assume fixed recourse and as a consequence the values and dimensions of the recourse matrix can be uncertain. The proposed algorithm contains multi-cut (partial) Benders decomposition and the deterministic equivalent model as special cases and can be used to trade-off computational speed and memory requirements. The algorithm outperforms multi-cut (partial) Benders decomposition in computational time and the deterministic equivalent model in memory requirements for a maintenance location routing problem. In addition, for instances with a large number of scenarios, the algorithm outperforms the deterministic equivalent model in both computational time and memory requirements. Furthermore, we present an adaptive relative tolerance for instances for which the solution time of the master problem is the bottleneck and the slave problems can be solved relatively efficiently. The adaptive relative tolerance is large in early iterations and converges to zero for the final iteration(s) of the algorithm. The combination of this relative adaptive tolerance with the proposed algorithm decreases the computational time of our instances even further.

To discuss the application of Internet of Things (IoT) in the agriculture, an agricultural product price prediction model is constructed based on the improved Elman neural network (ENN) of deep learning. Simulation experiment of pest prediction is carried out based on MATLAB, and then the agricultural IoT information service system is combined with the improved ENN prediction model to predict and analyze agricultural product prices. The results show that the agricultural product price prediction model based on the improved ENN has high accuracy, which reaches 0.9241. Through this agricultural product price prediction model, the agricultural IoT information service system can predict the price trend of agricultural products better, and it can also predict the shortcomings in the production and sales of agricultural products, so that corresponding management and control measures can be realized. This work can provide an important reference for the application of deep learning methods in agricultural IoT.

Connected and automated vehicle (CAV) technology is providing urban transportation managers tremendous opportunities for better operation of urban mobility systems. However, there are significant challenges in real-time implementation as the computational time of the corresponding operations optimization model increases exponentially with increasing vehicle numbers. Following the companion paper (Chen et al. 2021), which proposes a novel automated traffic control scheme for isolated intersections, this study proposes a network-level, real-time traffic control framework for CAVs on grid networks. The proposed framework integrates a rhythmic control method with an online routing algorithm to realize collision-free control of all CAVs on a network and achieve superior performance in average vehicle delay, network traffic throughput, and computational scalability. Specifically, we construct a preset network rhythm that all CAVs can follow to move on the network and avoid collisions at all intersections. Based on the network rhythm, we then formulate online routing for the CAVs as a mixed integer linear program, which optimizes the entry times of CAVs at all entrances of the network and their time–space routings in real time. We provide a sufficient condition that the linear programming relaxation of the online routing model yields an optimal integer solution. Extensive numerical tests are conducted to show the performance of the proposed operations management framework under various scenarios. It is illustrated that the framework is capable of achieving negligible delays and increased network throughput. Furthermore, the computational time results are also promising. The CPU time for solving a collision-free control optimization problem with 2,000 vehicles is only 0.3 second on an ordinary personal computer.

We study how delivery data can be applied to improve the on-time performance of last-mile delivery services. Motivated by the delivery operations and data of a food delivery service provider, we discuss a framework that integrates travel-time predictors with order-assignment optimization. Such integration enables us to capture the driver’s routing behavior in practice as the driver’s decision-making process is often unobservable or intricate to model. Focusing on the order-assignment problem as an example, we discuss the classes of tractable predictors and prediction models that are highly compatible with the existing stochastic and robust optimization tools. We further provide reformulations of the integrated models, which can be efficiently solved with the proposed branch-and-price algorithm. Moreover, we propose two simple heuristics for the multiperiod order-assignment problem, and they are built upon single-period solutions. Using the delivery data, our numerical experiments on a real-world application not only demonstrate the superior performance of our proposed order-assignment models with travel-time predictors, but also highlight the importance of learning behavioral aspects from operational data. We find that a large sample size does not necessarily compensate for the misspecification of the driver’s routing behavior. This paper was accepted by Hamid Nazerzadeh, big data analytics.

In this study, we propose an integrated model of two‐sided stochastic matching platforms to understand the design and operations of free‐float shared‐mobility systems. In particular, we address the joint design of incentives (via “crowd‐sourcing”) and spatial capacity allocations (enabled by “geo‐fencing”). From the platform's perspective, we formulate stylized models based on strategic double‐ended queues. We optimize the design and operations of such systems in a case study using a data set from a leading free‐float bicycle‐sharing system, and solve it via mixed‐integer second‐order conic programs (SOCPs). Both stylized results and computational studies generate insights about fundamental trade‐offs and triangular relationships among operational costs, capacity utilization rates and service levels. Interestingly, we identify the role of spatial capacity (parking) management to fine‐tune the market thickness (transient service availability) in such a two‐sided marketplace. We show that a “capacity‐dependent approximation” can be very close to optimality, and outperforms policies ignoring capacity management. We also demonstrate that this framework can be operationalized in multiple directions, which generates insights concerning matching efficiency, performance comparison between crowd‐sourcing and repositioning, strategic server behaviors and network externalities. Our insights guide the platform and the policy‐maker to embrace “crowd‐sourcing” and “geo‐fencing” technologies for shared‐mobility systems.

Data center networks provide the physical infrastructure that hosts Internet‐related services and cloud computing. Designing data center networks properly is imperative for Internet‐related service and cloud computing providers to gain competitive edges through cost efficiency and service quality. In this study, we formulate a mathematical programming model to address the data center network design problem, in which the objective is to minimize total operating cost and the service delay penalty by optimizing data center location, footprint allocation, and resource provisioning decisions, while incorporating essential features, such as latency, power, multiple resources, configuration limits, and interdependent footprints. We employ a queueing model to approximate the service latency and provide tractable reformulations. To enhance computational efficiency for large‐scale problems, we further develop Lagrangian relaxation methods and generate strengthening cuts by exploiting the structural properties of the problem. Our numerical studies demonstrate that the proposed model, which jointly optimizes location, allocation, and resource provisioning, can achieve significant cost reductions and improvements in service quality compared with a hierarchical approach that optimizes these decisions sequentially. Moreover, our proposed solution methods outperform state‐of‐the‐art commercial software in terms of computational efficiency. Based on real‐world datasets, the proposed model selects data centers that have been chosen by major cloud computing infrastructure providers. We also draw managerial insights that can be used as design guidelines in practice.

In recent years, the frequency, intensity, and severity of wildfires have been on the rise due to various environmental factors. Several studies show that the strategic application of fuel treatments is effective at altering fire behavior and its spread patterns. Effective planning for mitigating future expected losses under wildfire risk is a complex challenge that requires the integration of fire spread, simulation, and optimization models as well as the inclusion of multiple objectives into a unified framework. Previous works simplify the analysis by valuing the landscape regions using a unique objective (e.g., minimize the average expected area burned) or a predefined objective function. However, such an assumption is a simplification of the real system as multiple parts of the landscape have different values based on factors such as the presence of human settlements and infrastructure, availability of environmental services, and forest health. In this work, we expand these previous attempts by providing an integrated framework to naturally include and weight multiple objectives into the optimization model and analyze the trade-off between present objectives and future protection against wildfire risk. We study three key regions based on their recent fire history, landscape diversity, and demographic variety to quantify the impact of multiple objectives in landscape management. We obtain treatment plans using various combinations of these layers reflecting how different priorities of the decision-makers could affect treatment policies.

Apiculture has gained worldwide interest because of its contributions to economic incomes and environmental conservation. In view of these, migratory beekeeping, as a high-yielding technique, is extensively adopted. However, because of the lack of an overall routing plan, beekeepers who follow the experiential migratory routes frequently encounter unexpected detours and suffer losses when faced with problems such as those related to nectar source capacities and the production of bee products. The migratory beekeeping routing problem (MBRP) is proposed based on the practical background of the commercial apiculture industry to optimize the global revenue for beekeepers by comprehensively considering nectar source allocation, migration, production and sales of bee products, and corresponding time decisions. The MBRP is a new variant of the vehicle routing problem but with significantly different production time decisions at the vertices (i.e., nectar sources). That is, only the overlaps between residence durations and flowering periods generate production benefits. Different sales visits cause different gains from the same products; in turn, these lead to different production time decisions at previously visited nectar source locations and even change the visits for production. To overcome the difficulty resulting from the complicated time decisions, we utilize the Dantzig–Wolfe decomposition method and propose a revised labeling algorithm for the pricing subproblems. The tests, performed on instances and a real-world case, demonstrate that the column generation method with the revised labeling algorithm is efficient for solving the MBRP. Compared with traditional routes, a more efficient overall routing schedule for migratory beekeepers is proposed. Summary of Contribution. Based on the practical background of commercial apiculture industry, this paper proposes a new type of routing problem named the migratory beekeeping routing problem (MBRP), which incorporates the selection of productive nodes and sales nodes as well as the production time decision at the productive nodes on a migratory beekeeping network. To overcome the difficulty resulting from the complicated time decisions, we utilize the Dantzig–Wolfe decomposition method and propose a revised labeling algorithm for the pricing subproblems. The tests, performed on instances and a real-world case, demonstrate that the column generation method with the revised labeling algorithm is efficient for solving the MBRP. Compared with traditional routes, a more efficient overall routing schedule for migratory beekeepers is proposed. Therefore, this paper is congruent with, and contributes to, the scope and mission of INFORMS Journal on Computing, especially the area of Network Optimization: Algorithms & Applications.

A robotic mobile fulfillment system (RMFS) performs the order fulfillment process by bringing inventory to workers at pick-pack-and-ship warehouses. In the RMFS, robots lift and carry shelving units, called inventory pods, from storage locations to picking stations where workers pick items off the pods and put them into shipping cartons. The robots then return the pods to the storage area and transport other pods. In this article, we consider an item assignment problem in the RMFS in order to maximize the sum of similarity values of items in each pod. We especially focus on a reoptimization heuristic to address the situation where the similarity values are altered so that a good assignment solution can be obtained quickly with the changed similarity values. A constructive heuristic algorithm for the item assignment problem is developed, and then, a reoptimization heuristic is proposed based on the constructive heuristic algorithm. Then, computational results for several instances of the problem with 10–500 items are presented. We further analyze the case for which an item type can be placed into two pods. Note to Practitioners —This article proposes an efficient heuristic algorithm for assigning items to pods in a robotic mobile fulfillment system (RMFS) so that items ordered together frequently are put into the same pod. Computational results with 10–500 items show that the gaps from upper bounds are very small on average. For cases where the similarity values between items change or their estimation is not accurate due to the fluctuations in demand, a reoptimization heuristic algorithm that alters the original assignment is developed. The experimental results show that the reoptimization algorithm is robust when perturbation levels are approximately 40%–50% of the original similarity values with much less computation times. We believe that this research work can be very helpful for operating the RMFS efficiently.

As recommender systems send a massive amount of content to keep users engaged, users may experience fatigue which is contributed by 1) an overexposure to irrelevant content, 2) boredom from seeing too many similar recommendations. To address this problem, we consider an online learning setting where a platform learns a policy to recommend content that takes user fatigue into account. We propose an extension of the Dependent Click Model (DCM) to describe users' behavior. We stipulate that for each piece of content, its attractiveness to a user depends on its intrinsic relevance and a discount factor which measures how many similar contents have been shown. Users view the recommended content sequentially and click on the ones that they find attractive. Users may leave the platform at any time, and the probability of exiting is higher when they do not like the content. Based on user's feedback, the platform learns the relevance of the underlying content as well as the discounting effect due to content fatigue. We refer to this learning task as “fatigue-aware DCM Bandit” problem. We consider two learning scenarios depending on whether the discounting effect is known. For each scenario, we propose a learning algorithm which simultaneously explores and exploits, and characterize its regret bound.

Most pricing and revenue management models have at their core an optimization problem; one needs to determine the optimal price or quantity to maximize a profit or revenue function. To ensure tractability, conditions that assure the objective function has a unique solution are enormously helpful. So far, several technical assumptions have been proposed for the continuous case, but comparatively little attention has been given to the discrete counterpart despite its prevalence in practice. Thus, this study aims to develop new technical assumptions, built upon relevant economic concepts, to guarantee the tractability of revenue management models in discrete settings. In particular, we present two sufficient conditions for the revenue function to be concave, in terms of quantity or price and propose a condition for the revenue function to be unimodal, called discrete increasing generalized failure rate (IGFR). Our definition has an appropriate economic interpretation and offers comparable properties to those of the continuous version. Finally, we show the discrete IGFR property holds for several discrete distributions.

In this paper, a single-step framework for predicting quantiles of time series is presented. Subsequently, we propose that this technique can be adopted as a data-driven approach to determine stock levels in the environment of newsvendor problem and its multi-period extension. Theoretical and empirical findings suggest that our method is effective at modeling both weakly stationary and some nonstationary time series. On both simulated and real-world datasets, the proposed approach outperforms existing statistical methods and yields good newsvendor solutions.

Electric vehicles (EVs) are widely considered to be a solution to the problems of increasing carbon emissions and dependence on fossil fuels. However, the adoption of EVs remains sluggish due to range anxiety, long charging times, and inconvenient and insufficient charging infrastructure. Various problems with EV service operations should be addressed to overcome these challenges. This study reviews the state-of-the-art mathematical modeling-based literature on EV operations management. The literature is classified according to recurring themes, such as EV charging infrastructure planning, EV charging operations, and public policy and business models. In each theme, typical optimization models and algorithms proposed in previous studies are surveyed. The review concludes with a discussion of several possible questions for future research on EV service operations management.

With the applications of electric buses (e-buses), potential solutions to problems related to infrastructures for charging e-buses are emerging. This study particularly focused on large-scale fast charging-station planning for e-buses in the public transportation electrification process, according to the characteristics of e-bus operation and plug-in fast charging mode. We conducted an interdisciplinary study to optimize planning jointly under the transportation system and power grid. In addition to capturing the spatiality of the e-bus charging service network, we further considered temporality in order to conduct long-term planning in view of the continuously growing e-bus charging demand. A spatial-temporal model, which determines the sites and sizes of e-bus charging stations, was proposed and the strategies for multistage infrastructure planning were put forward. The model was equivalently transformed into a mixed-integer second-order cone programming with high computational efficiency. The model and the multistage planning strategies were justified through a series of numerical experiments. A case study of Shenzhen, China was implemented and the robustness of the model to plan changes was studied.

In this paper, we develop an iterative piecewise linear approximation approach with a novel initialization method to solve natural gas pipeline transmission problems with the nonuniform network elevation. Previous approaches, such as energy minimization methods, cannot be applied to solve problems with the nonuniform network elevation because they exclude pressure range constraints, and thus provide solutions far from optimum. We propose a new initialization model that considers pressure range constraints and improves the optimality of the solutions and the computational efficiency. Furthermore, we extend the energy minimization methods and provide the necessary conditions under which the extended methods operate in networks with the nonuniform elevation. We test the performances of the methods with previously reported pipeline networks from the literature, with the open data set GasLib, and with our industrial collaborator. The initialization approach is shown to be more efficient than the method with fixed initial breakpoints. The newly proposed initialization approach generates solutions with a higher accuracy than the extended energy minimization methods, especially in large‐size networks. The proposed method has been applied to natural gas transmission planning by the China National Petroleum Corporation and has brought a direct profit increase of 330 million U.S. dollars in 2015‐2017.

This study is devoted to designing locations and capacities of charging stations for supporting long-distance travel by electric vehicles (EVs). We first establish an expanded network structure to model the set of valid charging strategies for EV drivers, and then a variational inequality (VI) is formulated to capture the equilibrated route-choice and charging behaviors of EVs by incorporating an approximated queuing time function for a capacitated charging facility. Next, we formulate the problem of designing the locations and capacities of charging facilities under a fixed budget constraint and solve the optimization problem with a customized neighborhood search strategy. A lower bound for the system cost is also developed to evaluate the qualities of solutions acquired using our proposed heuristic. Numerical examples with a toy network and a highway network extracted from the Yangtze River Delta are used to show the effectiveness of the proposed methodology, and we observe that our strategy can solve a large-scale problem within an optimality gap of less than 5%.

The β-robust machine scheduling has attracted increasing attention as an effective method to hedge against uncertainty. However, existing β-robust scheduling models rely on the normality assumption of uncertain parameters, and existing solution methods are based on branch and bound, which cannot solve problems of 45 jobs within 3,600 seconds. This paper proposes distributionally β-robust scheduling (DRS) models to handle uncertain processing times. The DRS models only require the lower bound, mean, and covariance information of processing times, and have the capability of handling both single and parallel machine problems. Another key contribution of this paper is to devise efficient parametric search (PS) methods for the DRS models. Specifically, we show that there exists a parameterized assignment problem (PAP), such that its optimal solutions are also optimal for the original problem. The proposed methods only need to perform a one-dimensional PS and solve a series of PAPs. We further propose a bidirectional PS to reduce the number of PAPs needed to be solved, and we design a speedup shortest augmentation path algorithm for these PAPs. Experimental results on both single and identical parallel machine problems show that the improved PS method outperforms existing algorithms by more than three orders of magnitude improvement in computation time for problems of 45 jobs, and it is able to solve problems of 500 jobs within 0.5 seconds.

It is well‐known that adding a little flexibility to the right place is an effective strategy to improve the performance of operations in the face of demand uncertainties, to ensure high level of capacity utilization. However, given that system disruptions are ubiquitous, the legacy flexibility designs may perform poorly under disruptions to supply or capacity installations. In this study, we focus on the design of reliable and sparse flexibility structures that consistently meet a reasonable performance criterion under disruptions to both demand and supply. Specifically, we propose a class of structures termed as extended probabilistic expanders, based on the conjecture that the expansion property, rather than the global connectivity, is critical to good performance of the structures. We prove that for a system with n retailers, essentially only O(n) supply routes between suppliers and retailers are necessary to ensure good performance under disruption. In addition, we present an efficient randomized algorithm to construct extended probabilistic expanders, and demonstrate that the construction yields very good structure with the least number of edges asymptotically. We also investigate an extension to systems with structural constraints. Numerical results demonstrate that our design has not only a wide range of applications, but also better performance than a variety of known structures.

We herein investigate the single item, multi-period inventory management problem with fixed ordering cost, holding cost, and lost sales cost. Here, both the demand and vendor lead time (VLT) are uncertain; Additionally, the contextual information that may be useful for predicting the distribution of the demand and VLT is available. We assume that the retailer has access to rich history observations and wishes to decide: (1) when to place an order and (2) how much to order. We solve this problem via an end-to-end (E2E) learning approach, where we develop a deep learning framework that directly outputs the optimal order timing and quantity with the given contextual information as the input. Our numerical experiments, which use the real-world data obtained from an online retailer, demonstrate the advantages of our approach over benchmark methods, such as the (r, Q) policy, predict-then-optimize (PTO) framework, and the E2E method with given order timing. Furthermore, we analyze the convergence of our E2E approach and provide upper bounds for the expected daily average cost of our E2E approach under certain conditions. To compare our approach with the optimal policy, we also provide lower bounds for the cost of the optimal policy under certain assumptions.

Digital Twin, a new advanced digitalization paradigm, has attracted interest worldwide over the past five years. Practitioners and scholars alike still struggle with defining and realizing Digital Twin’s possibilities in real-life scenarios. To resolve such confusion, we highlight four essential elements of successful Digital Twin implementation, namely data, simulation, connection, and human. The four elements’ necessity is explained well through three proposed critical challenges: (1) digital at scale, (2) decision intelligence in data-intensive systems, and (3) consistency between objectives, decisions, and execution. We recognize that Operations Research (OR) expertise plays a vital role in building an efficient and intelligent Digital Twin. We identify six promising research opportunities for the OR community that could expand the scope and depth of OR theory.

Increasing testing capacities plays a substantial role in safely reopening the economy and avoiding a new wave of COVID-19. Pooled testing can expand testing capabilities by pooling multiple individual samples, but it also raises accuracy concerns. In this study, we propose a flexible testing strategy that adopts pooled testing or individual testing according to epidemic dynamics. We identify the prevalence threshold between individual and pooled testing by modeling the expected number of tests per confirmed case. Incorporating an epidemic model, we show pooled testing is more effective in containing epidemic outbreaks and can generate more reliable test results than individual testing because the reliability of test results is relevant to both testing methods and prevalence. Our study is the first to evaluate the interplay between pooled testing and a rapidly evolving outbreak to the best of our knowledge. Our results allay accuracy concerns about pooled testing and provide theoretical supports to empirical studies.

We review the operations research/management science literature on data‐driven methods in retail operations. This line of work has grown rapidly in recent years, thanks to the availability of high‐quality data, improvements in computing hardware, and parallel developments in machine learning methodologies. We survey state‐of‐the‐art studies in three core aspects of retail operations—assortment optimization, order fulfillment, and inventory management. We then conclude the paper by pointing out some interesting future research possibilities for our community.

Problem definition: In this paper, we study a reliable hub location model, the objective of which is to minimize the associated costs plus the penalty for unserved demands. The model assigns to eac...

Problem definition: Providing fast and reliable delivery services is key to running a successful online retail business. To achieve a better delivery time guarantee policy, we study how to estimate and promise delivery time for new customer orders in real time. Academic/practical relevance: Delivery time promising is critical to managing customer expectations and improving customer satisfaction. Simply overpromising or underpromising is undesirable because of the negative impacts on short-/long-term sales. To the best of our knowledge, we are the first to develop a data-driven framework to predict the distribution of order delivery time and set promised delivery time to customers in a cost-effective way. Methodology: We apply and extend tree-based models to generate distributional forecasts by exploiting the complicated relationship between delivery time and relevant operational predictors. To account for the cost-sensitive decision-making problem structure, we develop a new split rule for quantile regression forests that incorporates an asymmetric loss function in split point selection. We further propose a cost-sensitive decision rule to decide the promised delivery day from the predicted distribution. Results: Our decision rule is proven to be optimal given certain cost structures. Tested on a real-world data set shared from JD.com, our proposed machine learning–based models deliver superior forecasting performance. In addition, we demonstrate that our framework has the potential to provide better promised delivery time in terms of sales, cost, and accuracy as compared with the conventional promised time set by JD.com. Specifically, our simulation results indicate that the proposed delivery time promise policy can improve the sales volume by 6.1% over the current policy. Managerial implications: Through a more accurate estimation of the delivery time distribution, online retailers can strategically set the promised time to maximize customer satisfaction and boost sales. Our data-driven framework reveals the importance of modeling fulfillment operations in delivery time forecasting and integrating the decision-making problem structure with the forecasting model.

At the end of 2017, the leading ride-hailing platform DiDi Express replaced the commonly used surge pricing mechanism with a queueing mechanism, called the virtual queueing mechanism, in a number of cities across China to mitigate public relations pressure. To explore the benefits of the virtual queueing mechanism over the surge pricing mechanism in terms of operational efficiency, equity, and consumer welfare, we build a model of a data-driven state-dependent open queueing network, which we show can be approximated by an M/M/s+M queueing model with balking and reneging. We derive closed-form expressions for multiple performance metrics, including response rate, demand satisfaction rate, gross merchandise value (GMV), consumer surplus, and degree of inequality. Our theoretical and numerical analyses show that: (1) the virtual queueing mechanism outperforms the surge pricing mechanism in terms of consumer surplus, but is not as advantageous to generate GMV; (2) with respect to the response rate and demand satisfaction rate, the virtual queueing mechanism performs increasingly better than the surge pricing mechanism as demand increases; (3) the virtual queueing mechanism contributes to a more equitable ridesharing system, in the sense that consumer surplus is distributed across different classes of riders in a more balanced way. Using real data, we conduct a case study in Beijing, which verifies all the aforementioned results. The case study further finds that the response and demand satisfaction rates are significantly correlated with driver idleness.

As digitization enables service platforms to access users' information, important questions arise about how digital service platforms should disseminate information to improve platform operations. We contribute to this understanding by examining how providing customer-related information at the beginning of a service encounter affects service capacity and enjoyment. We evaluate this strategy via a field experiment on a live-streaming platform that connects hundreds of millions of individual broadcasters and viewers. When viewers entered shows, we provided viewer-related information to broadcasters who were randomly assigned to the treatment condition (but not to control broadcasters). Our analysis, involving a subsample of 49,998 broadcasters, demonstrates that relative to control broadcasters, treatment broadcasters expanded service capacity by 12.62% by increasing both show frequency (3.31%) and show length (7.10%), thus earning 10.44% more based on our conservative estimate. Moreover, our intervention increased service enjoyment (measured by viewer watch time) by 4.51%. Two surveys and additional analyses provide evidence for two mechanisms: (1) information about viewers presented shortly after their entry enables personalized greetings and facilitates social interaction, which addresses both viewers' and broadcasters’ needs for affiliation and satisfies viewers' preference for personalized treatment; (2) broadcasters receiving viewer-related information have a more vivid depiction of viewers and thus feel that their service is more appreciated and meaningful, which addresses broadcasters’ needs for competence. We also rule out several alternative explanations. Our low-cost, information-based intervention has important implications for digital service platforms that have little control over service providers’ work schedules and service quality.

We study a novel practice in the urban logistics sector, the transshipment-based urban courier (TUC) system. In the TUC system, large-capacity vehicles are utilized as moving warehouses to flexibly consolidate packages from individual couriers. The goal is to achieve a better trade-off between transportation costs and delivery efficiency. Because the performance of the TUC system depends significantly on the coordination of the transportation services and the pricing system for consumers, we consider the joint transportation coverage and pricing problem. In particular, we propose a data-driven robust pricing model in coordination with a transportation coverage model. Theoretical performance guarantees and numerical results verify the robustness of the pricing decision. The coverage-and-pricing model can be solved by a decomposition-based heuristic. This heuristic is a valid approximation algorithm under sufficient regularity conditions. After decomposition, the coverage-and-pricing problem can be reformulated as a maximum weighted coverage problem and solved via column generation, where the column-generating subproblems can be solved efficiently by a dynamic programming algorithm. Numerical experiments and a case study based on operational data from DML-Express demonstrate the efficiency of our method and provide managerial implications for the pricing policy in a TUC system.

We study an urban bike lane planning problem based on the fine-grained bike trajectory data, which is made available by smart city infrastructure such as bike-sharing systems. The key decision is where to build bike lanes in the existing road network. As bike-sharing systems become widespread in the metropolitan areas over the world, bike lanes are being planned and constructed by many municipal governments to promote cycling and protect cyclists. Traditional bike lane planning approaches often rely on surveys and heuristics. We develop a general and novel optimization framework to guide the bike lane planning from bike trajectories. We formalize the bike lane planning problem in view of the cyclists' utility functions and derive an integer optimization model to maximize the utility. To capture cyclists' route choices, we develop a bilevel program based on the Multinomial Logit model. We derive structural properties about the base model and prove that the Lagrangian dual of the bike lane planning model is polynomial-time solvable. Furthermore, we reformulate the route choice based planning model as a mixed integer linear program using a linear approximation scheme. We develop tractable formulations and efficient algorithms to solve the large-scale optimization problem. Via a real-world case study with a city government, we demonstrate the efficiency of the proposed algorithms and quantify the trade-off between the coverage of bike trips and continuity of bike lanes. We show how the network topology evolves according to the utility functions and highlight the importance of understanding cyclists' route choices. The proposed framework drives the data-driven urban planning scheme in smart city operations management.

Problem definition: We consider intertemporal pricing in the presence of reference effects and consumer heterogeneity. Our research question encompasses how to estimate heterogeneous consumer reference effects from data and how to efficiently compute the optimal pricing policy. Academic/practical relevance: Understanding reference effects is essential for designing pricing policies in modern retailing. Our work contributes to this area by incorporating consumer heterogeneity under arbitrary distributions. Methodology: We propose a mixed logit demand model that allows arbitrary joint distributions of valuations, responsiveness to prices, and responsiveness to reference prices among consumers. We use a nonparametric estimation method to learn consumer heterogeneity from transaction data. Further, we formulate the pricing optimization as an infinite horizon dynamic programming problem and solve it by applying a modified policy iteration algorithm. Results: Moreover, we investigate the structure of optimal pricing policies and prove the suboptimality of constant pricing policies even when all consumers are loss-averse according to the classical definition. Our numerical studies show that our estimation and optimization framework improves the expected revenue of retailers via accounting for heterogeneity. We validate our model using real data from JD.com, a large E-commerce retailer, and find empirical evidence of consumer heterogeneity. Managerial implications: In practice, ignoring consumer heterogeneity may lead to a significant loss of revenue. Furthermore, heterogeneous reference effect offers a strong motive for promotions and price fluctuations.

This work focuses on solving the data-driven contextual newsvendor problem with intertemporal dependence and non-stationarities. More specifically, we investigate learn the data-to-decision mapping for the newsvendor problem when observations of contexts and demands are available. The observations of both contexts and demands are generated sequentially in a fluctuate nature, thus exhibit an intertemporal dependence and even non-stationarities. However, most existing works that investigate the data-driven conditional Newsvendor problem adopt a common assumption that the data are independent and identically distributed (i.i.d.) to obtain performance guarantees such as generalization bounds. In this work, we develop performance guarantees in the form of out-of-sample generalization bounds for learning contextual newsvendor problem under comparatively more realistic assumptions including intertemporal dependence and moderate non-stationarities.

We investigate a data-driven multiperiod inventory replenishment problem with uncertain demand and vendor lead time (VLT) with accessibility to a large quantity of historical data. Different from the traditional two-step predict-then-optimize (PTO) solution framework, we propose a one-step end-to-end (E2E) framework that uses deep learning models to output the suggested replenishment amount directly from input features without any intermediate step. The E2E model is trained to capture the behavior of the optimal dynamic programming solution under historical observations without any prior assumptions on the distributions of the demand and the VLT. By conducting a series of thorough numerical experiments using real data from one of the leading e-commerce companies, we demonstrate the advantages of the proposed E2E model over conventional PTO frameworks. We also conduct a field experiment with JD.com, and the results show that our new algorithm reduces holding cost, stockout cost, total inventory cost, and turnover rate substantially compared with JD’s current practice. For the supply chain management industry, our E2E model shortens the decision process and provides an automatic inventory management solution with the possibility to generalize and scale. The concept of E2E, which uses the input information directly for the ultimate goal, can also be useful in practice for other supply chain management circumstances. This paper was accepted by Hamid Nazerzadeh, big data analytics. Funding: This research was supported by the National Key Research and Development Program of China [Grant 2018YFB1700600] and National Natural Science Foundation of China [Grants 71991462 and 91746210]. Supplemental Material: The online data are available at https://doi.org/10.1287/mnsc.2022.4564 .

We address a dynamic covering location problem of an unmanned aerial vehicle base station (UAV-BS), in which the location sequence of a single UAV-BS in a wireless communication network is determined to satisfy data demand arising from ground users. This problem is especially relevant in the context of smart grid and disaster relief. The vertical movement ability of the UAV-BS and nonconvex covering functions in wireless communication restrict utilizing classical planar covering location approaches. Therefore, we develop new formulations to this emerging problem for a finite time horizon to maximize the total coverage. In particular, we develop a mixed-integer nonlinear programming formulation that is nonconvex in nature and propose a Lagrangean decomposition algorithm (LDA) to solve this formulation. Because of the high complexity of the problem, the LDA is still unable to find good local solutions to large-scale problems. Therefore, we develop a continuum approximation (CA) model and show that CA would be a promising approach in terms of both computational time and solution accuracy. Our numerical study also shows that the CA model can be a remedy to build efficient initial solutions for exact solution algorithms. Summary of Contribution: This paper addresses a facet of mixed integer nonlinear programming formulations. Dynamic facility location problems (DFLPs) arise in a wide range of applications. However, classical DFLPs typically focus on the two-dimensional spaces. Emerging technologies in wireless communication and some other promising application areas, such as smart grids, have brought new location problems that cannot be solved with classical approaches. For practical reasons, many research attempts to solve this new problem, especially by researchers whose primary research area is not OR, have seemed far from analyzing the characteristics of the formulations. Rather, solution-oriented greedy heuristics have been proposed. This paper has two main objectives: (i) to close the gap between practical and theoretical sides of this new problem with the help of current knowledge that OR possesses to solve facility location problems and (ii) to support the findings with an exhaustive computational study to show how these findings can be applied to practice.

A flow picking system in which the existing picking list is updated in real time has been considered as an effective solution for e-commerce warehouses to increase order fulfillment efficiency. The pivotal issues of performance analysis of flow picking systems, and comparison between batch picking systems and flow picking systems are of great concern, both for academics and practitioners of warehouse operation management. In this study, we first develop analytic models to estimate the critical performance indicators of a flow picking system, including picking density and turnover time of an order. Second, we leverage the proposed models and real warehouse data to compare the performance of batch picking and flow picking systems through simulation. Our results show that a flow picking system requires fewer order pickers and shorter walking distances than a batch picking system in most scenarios, especially those with a higher order arrival rate to achieve the same service level. Our study can provide valuable guidelines to warehouse managers and decision-makers for choosing an order fulfillment solution by comparing a batch picking system and a flow picking system.

With cloud robotics, particularly robotic vision available wi-thin a household, human are able to live a convenient and safer life in an ambient assisted living environment. Recent advances in computational intelligence including neural network improves the computational capability of the robotic vision to better understand the environment. Recently, internet hoaxes that affected the social community greatly have raised strong awareness among public in parental control and the content that the youngster can view. Therefore, this paper focuses on filtering movies or videos that is not suitable for youngster by attempting to identify movie genre. Movie genre classification has been investigated in recent years, but there exist noise in normal videos referred as generic frames, as mentioned in [1], that makes differentiation movies with similar frame difficult. A filtering approach is proposed in this paper in order to identify generic frames within the video and discard them from genre classification process, in order to improve genre classification performance. Experiment shows that the filtering approach are able to improve action genre class, but have difficulties and improving other genre classes.

Problem definition: Nonrenewable energy is typically recognized as ''dirty energy'' with respect to the environment such that most countries reduce their nonrenewable facilities and then improve their renewable energy adoption. In this paper, we study and develop effective hybrid energy generation and charging operations strategies with the aim of exploring and enhancing the environmental benefits of nonrenewable energy integrated electric vehicle (EV) charging service in a planning problem for photovoltaic (PV) charging stations. Academic/Practical relevance: Our research contributes to the literature on and practice of sustainable operations management by developing an integrated stochastic robust planning model that allows us to explore its implications in three areas: (1) the environmental benefits of nonrenewable energy synthesis, (2) integrated modeling with processor-sharing queuing to optimize the EV charging service coupled with renewable energy power scheduling, (3) the performance of proposed demand response (DR) policies that incorporate the features of intermittent energy, such as unstable solar resources. Methodology: We characterize the fleet charging dynamics by adopting a processor-sharing queuing network in a mixed integer programming (MIP) integrated planning model. We model the uncertain solar resources in a worst-case expectation weighted by geometric-like probabilities obtained by processor-sharing queuing modeling. We then develop its linear equivalent form to approximate a tractable optimization model. We also adopt a multinomial logit (MNL) model to estimate customers’ charing utilities and the incentive payment across time slots in our integrated DR policy. Results: In a real-world case study of CAR2GO in Amsterdam, we find that nonrenewable energy integration improves clean energy utilization and provides a reliable and cost-effective EV charging service under current PV charging practice. The results also demonstrate the advantages of the proposed DR policy in incorporating uncertain solar resources on the reliability, profitability and environmental benefits of the EV charging service. Managerial implications: In their efforts to improve renewable energy planning, policy-makers should be more prudent about giving up nonrenewable energy. Certain reasonable features - synthesizing DR operations coupled with efficient energy scheduling - can alleviate the adverse impacts of intermittency and uncertainties that plague renewable energy and, in turn, gradually improve the quality of EV charging services.

We review the operations research/ management science literature on data-driven methods in retail operations. This line of work has grown rapidly in recent years, thanks to availability of high-quality data, improvements in computing hardware, and parallel developments in machine learning methodologies. We survey state-of-the-art studies in three core aspects of retail operations -- assortment optimization, order fulfillment, and inventory management. We then conclude the paper by pointing out some interesting future research possibilities for our community.

There have been rapidly growing applications using machine learning models for predictive analytics in Electronic Health Records (EHR) to improve the quality of hospital services and the efficiency of healthcare resource utilization. A fundamental and crucial step in developing such models is to convert medical codes in EHR to feature vectors. These medical codes are used to represent diagnoses or procedures. Their vector representations have a tremendous impact on the performance of machine learning models. Recently, some researchers have utilized representation learning methods from Natural Language Processing (NLP) to learn vector representations of medical codes. However, most previous approaches are unsupervised, i.e. the generation of medical code vectors is independent from prediction tasks. Thus, the obtained feature vectors may be inappropriate for a specific prediction task. Moreover, unsupervised methods often require a lot of samples to obtain reliable results, but most practical problems have very limited patient samples. In this paper, we develop a new method called Prediction Task Guided Health Record Aggregation (PTGHRA), which aggregates health records guided by prediction tasks, to construct training corpus for various representation learning models. Compared with unsupervised approaches, representation learning models integrated with PTGHRA yield a significant improvement in predictive capability of generated medical code vectors, especially for limited training samples.

Most previous work in addressing the adaptive routing problem in stochastic and time-dependent (STD) network has been focusing on developing parametric models to reflect the network dynamics and designing efficient algorithms to solve these models. However, strong assumptions need to be made in the models and some algorithms also suffer from the curse of dimensionality. In this paper, we examine the application of Reinforcement Learning as a non-parametric model-free method to solve the problem. Both the online Q learning method for discrete state space and the offline fitted Q iteration algorithm for continuous state space are discussed. With a small case study on a mid-sized network, we demonstrate the significant advantages of using Reinforcement Learning to solve for the optimal routing policy over traditional stochastic dynamic programming method. And the fitted Q iteration algorithm combined with tree-based function approximation is shown to outperform other methods especially during peak demand periods.

Most studies of process flexibility have considered a make-to-order setting, whereas in practice, production systems usually have a make-to-stock environment. In this article, we investigate the pr...

With smartphones sweeping the world, customers use much more data service than voice service nowadays. Because mobile data service occupies much more network capacity than voice, wireless network is frequently congested. We build a dynamic cell phone plan control model where the service provider can decide a subset of plans to open at the beginning of each period. New customers can only subscribe open plans. The dynamic plan control problem is a stochastic programming problem where the system state dimension is exponential in the number of plans and periods. In this paper, we analyze the case when the number of customers approaches positive infinity. In this limiting case, we transform the stochastic dynamic programming problem to an equivalent integer linear programming problem. In a case study, we analyze a large mobile service provider in Asia using 10,000 real individual-level data. Sensitivity analysis implies that a mobile service provider whose initial market share is low can benefit more from the dynamic data plan control.

Rolling stock needs regular maintenance in a maintenance facility. Rolling stock from different fleets are routed to maintenance facilities by interchanging the destinations of trains at common stations and by using empty drives. We consider the problem of locating maintenance facilities in a railway network under uncertain or changing line planning, fleet planning, and other uncertain factors. These uncertainties and changes are modeled by a discrete set of scenarios. We show that this new problem is NP-hard and provide a two-stage stochastic programming and a two-stage robust optimization formulation. The second-stage decision is a maintenance routing problem with similarity to a minimum cost-flow problem. We prove that the facility location decisions remain unchanged under a simplified routing problem, and this gives rise to an efficient mixed-integer programming (MIP) formulation. This result also allows us to find an efficient decomposition algorithm for the robust formulation based on scenario addition (SA). Computational work shows that our improved MIP formulation can efficiently solve instances of industrial size. SA improves the computational time for the robust formulation even further and can handle larger instances due to more efficient memory usage. Finally, we apply our algorithms on practical instances of the Netherlands Railways and give managerial insights.

In 2017, Shenzhen replaced all its buses with battery e-buses (electric buses) and has become the first all-e-bus city in the world. Systematic planning of the supporting charging infrastructure for the electrified bus transportation system is required. Considering the number of city e-buses and the land scarcity, large-scale bus charging stations were preferred and adopted by the city. Compared with other EVs (electric vehicles), e-buses have operational tasks and different charging behavior. Since large-scale electricity-consuming stations will result in an intense burden on the power grid, it is necessary to consider both the transportation network and the power grid when planning the charging infrastructure. A cost-minimization model to jointly determine the deployment of bus charging stations and a grid connection scheme was put forward, which is essentially a three-fold assignment model. The problem was formulated as a mixed-integer second-order cone programming model, and a &ldquo, No R&rdquo, algorithm was proposed to improve the computational speed further. Computational studies, including a case study of Shenzhen, were implemented and the impacts of EV technology advancements on the cost and the infrastructure layout were also investigated.