Predictive models are increasingly embedded in operational decision-making, yet standard explanation methods typically explain forecasts rather than the decisions those forecasts induce. This distinction is important in predict-then-optimize systems: large forecast changes may leave the optimizer's action unchanged, while small changes can alter the selected decision and its realized value. We propose Decision Value Attribution (DVA), a Shapley-based framework for attributing the value of a fixed prediction–optimization pipeline. The framework defines cooperative games whose payoff is the downstream decision value, allowing the players to be information sources, optimization or design parameters, or both. We present three variants: InfoDVA attributes value to features, DesignDVA attributes value to operational configurations, and Decision-Value Interactions (DVI) quantifies how information and design jointly create value. We further distinguish post-DVA, which evaluates decisions using realized outcomes, from pre-DVA, which evaluates decisions under the model's full prediction. This separation turns attribution into a decision-level diagnostic of whether the model's operational beliefs align with realized performance. The resulting attributions are expressed in the units of the operational objective and decompose the gain or loss relative to a baseline. Case studies in electricity storage arbitrage and emergency medical service coverage show that predictive explanations can be poor proxies for operational value, that DVA can guide targeted information-control interventions, and that optimization configurations determine when predictive information is decision-relevant.

Truck-drone delivery systems have been proposed for sustainable and economical last-mile distribution, especially in urban environments. To widen the service range, some works have recommended adding facilities, such as drone stations, considering the problem in discrete space by choosing from a predefined set. In this article, an evolutionary optimization approach to the design decision of where to locate drone stations in the continuous plane is introduced, modeled, and solved. Drone stations serve as facilities for storage, charging, and launching. A truck (or other land transport means) transports parcels to the drone stations from a depot and the drones launch from the stations and deliver the parcels to each customer. The objective is to determine the positions of the drone stations in 2-D space and establish the shortest fixed truck route from the depot through all the stations and returning to the depot. The problem is constrained by the radius of service for each drone and all customers must be served, if possible. We formulate the problem as a constrained nonlinear optimization problem and present two versions of an algorithm using particle swarm optimization (PSO) with a subordinate dynamic program. Computational results show that our approach achieves much better results than a standard commercial nonlinear solver in a similar amount of computational time for both maximizing coverage of customers and minimizing distance of the truck delivery route. A design case study concerning healthcare delivery throughout the Birmingham, Alabama (USA) metropolitan area is provided.

Deep Reinforcement Learning (DRL) has shown considerable promise in addressing complex sequential decision-making tasks across various fields, yet its integration within Operations Research (OR) remains limited despite clear methodological compatibility. This paper serves as a practical tutorial aimed at bridging this gap, specifically guiding simulation practitioners and researchers through the process of developing DRL environments using Python and the Gymnasium library. We outline the alignment between traditional simulation model components, such as state and action spaces, objective functions, and constraints, and their DRL counterparts. Using an inventory control scenario as an illustrative example, which is also available online through our GitHub repository, we detail the steps involved in designing, implementing, and integrating custom DRL environments with contemporary DRL algorithms.

Offshore oil and gas operations are crucial for energy supply but encounter logistic challenges due to their remote locations and harsh environments. Costly shutdowns in this industry make the supply of spare parts a critical operation. Mobile additive manufacturing (MAM) factories provide a flexible and sustainable solution by enabling on-demand and on-site production, particularly in high-mix, low-volume contexts like spare parts. However, geographically dispersed facilities, like offshore platforms, require strategic distribution decisions since MAM factories may not be located in situ. Drones offer a promising solution for delivering products in offshore settings. No prior work integrates drone delivery and distribution decisions with MAM factories. This study addresses that gap by introducing a supply chain where MAM factories produce spare parts for offshore platforms, with delivery via drones or trucks/ships. A mixed-integer linear programming model is developed to support tactical and operational decision-making. The model is solved using a matheuristic that is benchmarked against an exact and an adaptive large neighbourhood search (ALNS) method. Computational experience shows the matheuristic outperforms ALNS in solution quality and CPU times. The approach is illustrated by a case study inspired by Fieldmade's operations, a MAM servicing North Sea oil platforms. Results highlight the advantages of both MAM factory relocations and drone deliveries in reducing shipping times and meeting demand due dates.

Public sentiment can impact the implementation of public policies and even cause policy failure if public support is not received. Therefore, knowledge of public sentiment concerning new and emerging policies is critical for policymakers. During the coronavirus disease 2019 (COVID-19) pandemic, several precautionary measures have been suggested in an attempt to delay or mitigate the spread of the virus. This study presents a framework that applies natural language processing (NLP) techniques, such as sentiment and bigram analyses, to characterize the public sentiment on three prominent mitigation measures (mask wearing, social distancing, and quarantine) as shared by Twitter users in the United States. As part of the framework, we apply a bigram graph-based approach to visualize the most frequent topics in Twitter discussions during the COVID-19 pandemic. The objective is to provide insights into the most commonly discussed topics among Twitter users with similar demographic characteristics (e.g., age and gender). The sentiment and bigram analyses identified the most frequently discussed topics expressing both positive and negative sentiments among different age and gender groups. Discussions containing positive sentiment prevailed and revolved around the benefits of the measures and trust in the government, while the topics of negative sentiment involved conspiracy theories, skepticism, and distrust of government mandates. It is also notable that the discussions among people 19–29 and over 40 years old focus on government officials and political parties, benefits or inefficiency of mitigation measures, and conspiracy theories more often than other demographic groups. Our proposed approaches and results offer a novel and potentially valuable contribution to public policymakers.

Last mile delivery is an important and growing part of the supply chain that has a sizeable negative environmental impact. This paper considers more sustainable approaches to home-delivery that are pragmatic for many non-rural environments. We address the two-echelon, multi-trip, capacitated vehicle routing problem with home-delivery and optional self-pickup services using different combinations of drones, trucks, and electric-assisted bikes (i.e. e-bikes). In the proposed approach, parcels are transported from a depot to parcel lockers by either drones or trucks and are then delivered to customer locations by either e-bikes or trucks. The four approaches range from the most sustainable (drones then e-bikes) to partly green (drone then truck or truck then e-bike) to finally the traditional (truck then truck). We formulate a mathematical model that determines the vehicle routes to minimize the total cost, which consists of vehicle operational cost and operator wages. The four types of delivery networks are assessed and compared for both costs and emissions. Experimental results suggest that with a modest increase in total cost (as little as 13%), emission reductions of up to 92%, on average, can be achieved when using the greenest delivery strategy of drones and e-bikes. The other delivery options have varying tradeoffs between costs and environmental impact. The percentage of self-pickup customers is an influencing factor to consider when choosing the delivery strategy that best meets the organization’s budget and environmental goals, especially when using e-bikes in the second echelon.

Choosing an appropriate optimization algorithm is essential to achieving success in optimization challenges. Here we present a new evolutionary algorithm structure that utilizes a reinforcement learning-based agent aimed at addressing these issues. The agent employs a double deep q-network to choose a specific evolutionary operator based on feedback it receives from the environment during optimization. The algorithm's structure contains five single-objective evolutionary algorithm operators. This single-objective structure is transformed into a multi-objective one using the R2 indicator. This indicator serves two purposes within our structure: first, it renders the algorithm multi-objective, and second, provides a means to evaluate each algorithm's performance in each generation to facilitate constructing the reinforcement learning-based reward function. The proposed R2-reinforcement learning multi-objective evolutionary algorithm (R2-RLMOEA) is compared with six other multi-objective algorithms that are based on R2 indicators. These six algorithms include the operators used in R2-RLMOEA as well as an R2 indicator-based algorithm that randomly selects operators during optimization. We benchmark performance using the CEC09 functions, with performance measured by inverted generational distance and spacing. The R2-RLMOEA algorithm outperforms all other algorithms with strong statistical significance (p<0.001) when compared with the average spacing metric across all ten benchmarks.

We study a same-day delivery problem where customer orders arrive dynamically throughout the day and the service operator must determine, in real-time, whether to accept the orders and how to adjust the ongoing distribution plan. We develop a route-based Markov Decision Process and an efficient online policy to dynamically route a truck that can receive newly arrived orders along its route via drones dispatched from a depot. Numerical experiments show that our online policy has an average fill rate decrease of at most 20% over the perfect-information counterpart. Further, this online policy has a fill rate increase of up to 8% over a naïve greedy policy. We also show that drone resupply increases fill rates by up to 21% compared to a conventional truck-only resupply system. Computational times to make each decision are in the hundredths of a second, thus allowing real-time feedback to customers regarding their eligibility for same-day delivery.

The Vehicle Routing Problem with Release Dates and Drone Resupply consists of routing a fleet of trucks to deliver orders that arrive at a depot over time. During the delivery horizon, the trucks can return to the depot to collect newly arrived orders, or these orders can be resupplied to the trucks along their routes via drones dispatched from the depot. A Mixed-Integer Linear Programming (MILP) formulation is developed for the version of the problem where order arrival times at the depot (generally termed order release dates) are known beforehand. To address large-size instances, we devise a unified matheuristic approach that provides high-quality solutions for both the truck-and-drone and the truck-only versions of the problem. In this approach, truck routes are iteratively modified using a tabu search scheme, where a subordinate fast MILP model defines optimal loading operations (truck depot returns and drone resupplies) for promising truck routes. We perform extensive numerical experiments with instances of up to 100 customers. Results show the effectiveness of the matheuristic approach for solving both the truck-and-drone and the truck-only versions of the problem. We also show the benefits of drone resupply to reduce completion times and the number of times the trucks need to return to the depot to collect newly released orders. Furthermore, we provide several managerial insights regarding fleet utilization and consolidation.

Improving last mile delivery, in terms of both efficiency and sustainability, has recently been enhanced using truck/drone tandems. However, nearly all these approaches focus only on deliveries, while returns are ignored. We propose a new model for integrating both delivery and returns in the combined operation of a truck and a drone, which we term the Flying Sidekick Traveling Salesman Problem integrating Deliveries and Returns with Multiple Payloads (FSTSP-DR-MP). This approach is more sustainable than truck-only deliveries as the drones operate through battery power with no emissions and can move more directly than vehicles along a road network. We formulate the problem as a mixed-integer linear program to minimize the total service time of the system, where the truck and the drone can perform both delivery and pickup of parcels during a single sortie. Because drones have a capacity of multiple parcels, each can visit more than one customer per dispatch, increasing drone utilization and responsiveness to customer expectations regarding return services, along with greater environmental improvements. Small-size cases are solved exactly using the MILP implemented in the CPLEX Python API. Since the MILP is not practical for realistically sized cases, we propose a meta-heuristic derived from Variable Neighborhood Search (VNS), which iteratively builds truck and drone routes. We show that this works well for up to 100 customers, a typical number in last-mile logistics. We assess the trade-offs between the ratio of returns to deliveries and drone capacity to provide managerial insights to the benefits of this approach for sustainable last mile logistics. Computational experiments show that integrating delivery and return truck-drone operations significantly reduces total service time and truck travel time compared to both traditional delivery schemes (single truck) and the well-known FSTSP drone schemes (one truck and one drone) up to 36.2% and 22.9%, respectively, by exploiting the nature of each customer (delivery or return) to increase the number of stops of each drone sortie. We also conduct a comparison with multiple drones have a single payload under similar scenarios. The results demonstrate that our model outperforms this alternative with an average improvement of 3.9% in total service time. Our approach addresses an important gap in the literature by accommodating returns, which are ubiquitous in last mile logistics, as well as providing for improved drone utilization, sustainability, and cost-effectiveness.

This paper describes two novel approaches to cost estimation of manufactured products where a data set of similar products have known manufactured costs. The methods use the notion of piecewise functions and are (1) clustering and (2) splines. Cost drivers are typically a mixture of categorical and numeric data which complicates cost estimation. Both clustering and splines approaches can accommodate this. Through four case studies, we compare our approaches with the often-used regression models. Our results show that clustering especially offers promise in improving the accuracy of cost estimation. While clustering and splines are slightly more complex to develop from both a user and a computational perspective, our approaches are packaged in an open-source software. This paper is the first known to adapt and apply these two well-known mathematical approaches to manufacturing cost estimation.

Despite the rise of e-commerce and online channels in the retail industry, brick and mortar stores continue to attract a large volume of customers and operate under high competition. A key element of traditional retailing is the store layout. This chapter investigates the influence of the store layout configuration on sales and in store traffic in a grocery store environment. A bi-objective optimization approach, combined with data mining techniques, is developed to examine two conflicting objectives, namely, customer satisfaction and revenue maximization. A case study is undertaken with the participation of Turkey’s largest grocery store chain, Migros, to test and evaluate the proposed approach. The results after the implementation of the new layout show that an analytical approach can result in superior revenue for the store.

This paper presents a two-phase approach for solving the facility layout problem in a physical rehabilitation hospital. The first phase solves the block layout problem, where the relative location and size of the departments in the facility are determined. The model used in this phase is based on Space Syntax which offers a series of tools that can be used to analyze and quantify spatial relations that are useful when modeling block layouts. Two Space Syntax-based metrics are introduced to model proximity and ease of access in layout designs, critical qualities in health care settings. A tabu search algorithm based on a novel nested-bay encoding is used to find the block layouts. A set of test cases from a large provider of rehabilitation hospitals shows the ability of this approach to handle healthcare-specific design requirements. An important concern for physical rehabilitation hospitals, where a large portion of the patient population is especially vulnerable to infectious diseases, and which has gained greater attention due to the COVID pandemic, is infection control. The approach herein is more capable of addressing control of infectious disease than existing metrics by providing designers with more granular control of space separation. Results show that the Space Syntax approach provides powerful, but easy to use, modeling capabilities, and that the resulting block layouts are more realistic. The second phase model is a mixed integer program for constructing corridor networks on a block layout that minimize travel distance, number of intersections, and maximum traffic on a turn. Both models are configurable so that facility designers can generate different designs according to their goals by changing the model parameters.

The bat algorithm (BA) is a recent swarm intelligence algorithm which can be a powerful tool for numerical optimization. However, the BA and most of its variants rely on a quite elitist strategy where the bat with the current global best fitness guides the population’s search direction. This may result in a slow convergence rate and low accuracy when searching in complex problem spaces. In this chapter, a dimension-based best bat for BA is formed by integrating both the current global best bat and favorable search information of the other bats in different dimensions separately. This strategy is embedded in the BA to guide the population to fly toward better directions. The proposed algorithm is tested on the IEEE CEC 2017 benchmark suite and is compared with some related algorithms. The experimental results demonstrate the effectiveness and robustness of the proposed BA enhancement.

Presents the recipients of CIS society 2022 award winners.

This paper presents an approach using a combination of data-driven and analytical models to design the layout of empty container depots where top-lifters (TLs) are used. A method to determine the optimum number of blocks along with the number of driving lanes is proposed where the size of the blocks is specified by the number of rows, tiers, and bays. For estimating the effects of the design variables on the TL cycle time, formulas to calculate the expected travel distance of these vehicles are derived based on geometry and a Markov chain model is used to obtain the times of retrieval and placement by the TLs using data gathered in an empirical study from a typical empty container yard. Together, the total cycle time (travel, retrieval, and placement) is then used as the objective function to evaluate alternative container yard layout options. Numerical examples from a case study are provided to illustrate the layout design procedure and show its effectiveness and pragmatism.