Omega-international Journal of Management Science最新文献

Public transportation usage prediction is valuable for the sustainable development of transportation systems, particularly in crowded megacities. Machine learning technologies are of great interest for predicting public transportation usage. While these technologies outperform many other techniques, they suffer from limited interpretability. Explainable artificial intelligence (XAI) tools and techniques that offer post-hoc explanations of the obtained predictions are gaining popularity. This paper proposes an advanced tree-based ensemble algorithm for public transportation usage rate prediction. We aim to explain the predictions both with the most widely used technique of XAI, Shapley additive explanation (SHAP) and in the light of the rules presented. To predict the total public transportation usage, the proposed model combines all types of public transportation, categorized as ferry, railway, and bus, unlike most existing studies focusing on a single kind of public transport. Besides the sort of transportation, the day of the week, whether the day is special, and the daily ratio of passenger types were identified as model features for predicting the daily usage of each type of public transportation. We tested the proposed model using an open data set from Izmir City, Turkey. While the model had superior prediction performance, the explanations showed that the type of public transportation, weekday, and the ratio of full-fare passengers have the highest SHAP values, and the model features have many interactions. We also validated our results using an online data set showing Google search trends.

The multiple depot vehicle scheduling problem (MDVSP) is one of the most studied problems in public transport service planning. It consists of assigning buses to each timetabled trip while respecting vehicle availability at each depot. Although service quality, and especially reliability, is the core of most transport agencies, the MDVSP is more often than not solved solely in a cost-efficient way. This work introduces a data-driven model to the reliable MDVSP with stochastic travel time (R-MDVSP-STT). The reliability of a schedule is assessed and accounted for by propagating delays using the probability mass function of the travel time of each timetabled trip. We propose a heuristic branch-and-price algorithm to solve this problem and a labeling algorithm with a stochastic dominance criterion for the associated subproblems. The solutions obtained are compared based on three metrics — under normal and extraordinary circumstances. Computational results on real-life instances show that our method can efficiently find good trade-offs between operational costs and reliability, improving the reliability of the solutions with little cost increase.

The $\sigma$-$\mu$ efficiency methodology has been recently introduced for multicriteria evaluation problems, based on the framework of Stochastic Multi-Attribute Acceptability analysis (SMAA), to address the uncertainty in the performance of a set of decision alternatives. The methodology builds, iteratively, a set of Pareto-Koopmans efficiency frontiers, which are used to assess the alternatives with respect to their expected performance and its variability, measured across different scenarios for the weights of the evaluation criteria. This paper presents an improved algorithmic implementation of this methodology that provides results that are consistent with the Pareto dominance relation between the alternatives. The proposed approach is employed to evaluate the performance of a sample of European banks which participated in the European stress tests conducted by the European Banking Authority, over the last five years available (2017–2021). The performance and efficiency of the banks is analyzed using financial criteria along with environmental, social, and governance (ESG) factors. Results from comprehensive and disaggregated analysis reveal performance disparities among banks in financial and ESG factors, highlighting the influence of country-specific green policies and individual bank practices. Valuable for the banking sector and regulators, the findings help identify operational inefficiencies and propose areas for performance enhancement, operational improvement, and innovation, with a focus on green practices.

In the last decades, natural disasters, such as earthquakes and landslides, have occurred frequently, seriously threatening the safety of people’s lives and property. How emergency material is scheduled and delivered efficiently to the affected sites after a disaster has become a critical issue in emergency management. Current studies on emergency material scheduling mainly focus on truck or helicopter transport. Inspired by the success of employing drones in commercial logistics, this work investigates the emergency material scheduling issue based on the cooperative transport of drones, helicopters, and trucks. Specifically, this paper considers the limited transport capacity, road conditions in the early stage of the disaster rescue, and affected sites restricted by road conditions that can only be served by helicopters or drones. The studied problem is formulated as a mixed integer programming model, and a two-stage heuristic algorithm is developed to solve the model. For the proposed model, instances of different sizes are generated, and extensive experiments are performed to test the efficiency of the proposed algorithm. The comparison between the solutions obtained by the two-stage algorithm and Gurobi Solver for the small instances validates the effectiveness of the proposed heuristic algorithm. Experimental results for the larger instances show that the proposed two-stage algorithm can effectively solve the problem of emergency material scheduling. Sensitivity analysis of ten typical instances is performed to provide managerial insights. Finally, a case study of the Sichuan earthquake and the visualization of transport routes are presented. The model and solving approach proposed in this work can provide essential decision references for emergency management decision-making.

Considering the potential loss of decision-making power associated with share pledging, shareholders need to pledge appropriate share proportion based on their own shareholding situation. To assist shareholders in handling this decision-making challenge, we design a share pledging decision-making method with three-way decision, which fully considers the characteristics of share pledging. Firstly, we construct a novel calculation method of conditional probability based on left-tail systemic risk and share return data, which appropriately measures the probability of share price plummeting. Then, we propose a learning method of loss functions with shareholding proportion information, which considers pledged share proportions. To enhance the flexibility of the proposed method in handling practical share pledging issues, we discuss two scenarios including complete shareholding proportion information and limited shareholding proportion information. Furthermore, we design three-way proportion decision (TWPD) through introducing particle swarm optimization algorithm, which can determine the share proportion of pledge and redemption with minimizing the loss of decision-making power, and develop TWPD to sequential three-way proportion decision (STWPD) for solving multi-period share pledging problem. Finally, we apply the proposed method to deal with share pledging problem of a Chinese firm based on real data collected from China Stock Market & Accounting Research Database, and provide implications for shareholders and firms in risk management during share pledging based on the results of comparison experiments and sensitivity analysis.

Overcrowding is a well-known major issue affecting the behavior of an Emergency Department (ED), as it is responsible for patients’ dissatisfaction and has a negative impact on the quality of workers’ performance. Dealing with overcrowding in an ED is complicated by lack of its precise definition and by exogenous and stochastic nature of requests to be served. In this paper, we present a Decision Support System (DSS) based on the integration of a Deep Neural Network for dealing with the sources of uncertainty and a simulation tool to evaluate how specific management policies affect the ED behavior. The DSS is designed to be run on-line, dynamically suggesting the most suitable policy to be implemented in the ED. We evaluate the performance of the DSS on a specific major ED located in northern Italy. Numerical results show that overcrowding can be considerably reduced by allowing a dynamic selection among a limited set of simple policies for queue management.

Supply chain disruption can occur for a variety of reasons, including natural disasters or market dynamics for which resilient strategies should be designed. If the disruption is profound and has dire consequences for the economy, it calls for the regulator’s intervention to minimize the impact for the betterment of the society. This paper investigates the minimum quota regulation on transport amounts of a shipping company with limited capacity that transports a group of products with heterogeneous transportation and production costs and prices. An interesting example can be found in the North American rail transportation market, where rail capacity is used for a variety of products and commodities, such as oil and grains. Similarly, in Europe, the supply chain for grain produced in Ukraine is disrupted by the Ukraine war and the blockade of the maritime transport routes. This siege puts pressure on the rail transport capacity of Ukraine and its neighboring countries to the west, which needs to be shared to ship a variety of products, including grains, military, and humanitarian supplies. Such situations require the proper execution of government intervention for effective management of limited transport capacity to avoid rippling effects throughout the economy. We propose mathematical models and solutions for market players and the government in a Canadian case study. Subsequently, the conditions that justify government intervention are identified, and an algorithm is presented to obtain the optimum minimum quotas.

While the provision of Online Consultation Services (OCS) has brought convenience to patients, the original offline outpatient workflow may struggle to maintain efficiency. Hospital managers’ concerns lie on how to reconcile the operations of offline service with the provision of OCS and what are the latent impacts of OCS on the operational performances of healthcare system. The outpatient appointment scheduling is a crucial part of the workflow decision that affects both quality of offline services and the fragmented idle time that doctors can reply to their Follow-up patients (FUP) during working hours. To smooth the fluctuations in workload of OCS, we introduce a novel decision variable—the recommended maximum number of FUP that a doctor should serve during overtime hours (FUP-MN)—and propose an inter-day service shifting strategy that any requests for OCS exceeding the FUP-MN will be deferred to be served in the fragmented idle time of the next session. We formulate a stochastic programming model to search for the optimal combination of the appointment schedule and FUP-MN. Propositions are conducted to demonstrate the model is valid not only in coordinating the dual demand for online and offline services, but also in restricting the potential overflow of OCS requests. We provide a general solution approach and also develop a heuristic algorithm to quickly obtain optimal workflows at different OCS demand levels. The numerical experimental results indicate that an appropriate introduction of OCS can give the healthcare system a strong ability to resist the adverse effects of uncertainty.

Problems with multiple conflicting criteria are usually modeled by the methods proposed in the field of Multi-Criteria Decision Making (MCDM). In MCDM, one of the most important topics is the weighting of criteria. On the other hand, classification is employed in numerous real-world issues, like disease diagnosis. Diverse algorithms have been developed for this purpose. It is important to evaluate the classification performance in every problem by evaluating algorithms. The evaluation of algorithms includes several conflicting criteria; Therefore, it can be represented as an MCDM problem. We aim to develop a new weighting method that can be used for the classification problem with more than two classes, involve the risk of threatening human life, and consider minor features for different diseases in weighting. At present, none of the existing weighting methods fulfill these requirements. This research presents a new method called “Criteria Weighting Based on Confusion Matrix (CWBCM)” and our innovation is that, for the first time, all these gaps are filled with this method. This method calculates the exact importance of criteria using the confusion matrix in machine learning. The proposed method has been implemented on six different datasets of three diseases: COVID-19, thyroid, and diabetes, and compared with two common methods, Shannon and AHP. Two methods, TOPSIS and EDAS, were also used to rank the classifiers. Finally, the results show that our method is superior to the other two weighting methods in all critical factors and has unique features that other methods do not have.

With rapid advances in digitization, many critical processes in transportation, industries, and our daily life rely on sensor measurements. With time, however, the measurements may get gradually biased and their precision deteriorates, leading to an enhanced risk of major disruptions caused by false sensor measurements. All single sensor measurements are uncertain and deviate from the true value. To detect malfunctioning sensors early on, a set of recent measurements of each sensor has to be constantly cross-checked against the measurements of a given number of other sensors, i.e., sensors should form a diagnosable network.

In this article, we examine the intelligent positioning of safety-relevant sensors at railways such that the installed sensors can constantly cross-check each other and the number of the required sensors is minimized. The arising sensor positioning problem (SPP) belongs to the family of the coordinated set covering problems with two binary matrices: the choice of columns in one matrix implies the selection of specific columns and rows in the other matrix. We formulate an integer program, provide some formal analysis of the SPP and design a customized large neighborhood search metaheuristic RuM, which finds close-to-optimality solutions fast. In our computational experiments, we show that if we ignore the diagnosability requirement, the installed sensors cannot sufficiently cross-check each other in most cases. However, it costs only a few (or even no) additional sensors to ensure the diagnosability of the sensor network.