Concentrated vehicle arrivals at limited input/output points often cause severe queuing and prolonged waiting times in automated container terminals. This study addresses a preemptive single yard crane scheduling problem in a perpendicular-layout terminal. Preemption allows the crane to temporarily interrupt a container move, stage it at an intermediate slot, serve another request, and later complete the operation. To minimize a weighted sum of makespan and total operational delay (comprising vehicle waiting and storage delays), we formulate a mixed-integer linear programming model and develop an iterated greedy framework with four preemptive variants. Extensive computational experiments confirm that at least one preemptive variant consistently outperforms a non-preemptive baseline across all problem sizes and objectives. Under the balanced objective, preemptive scheduling reduces total vehicle waiting time by 7.4–61.2
We consider the following version of online bin packing. Arriving items of rational sizes in (0, 1] belong to a constant number of types denoted by 1,2,… ,k , and there are unbounded numbers of available bins of these types. The cost of a bin depends on its type, and the sequence of costs is monotonically increasing. Every item of type i can be packed into a bin of type whose index is at least i, and no bin may have a total size of items that is larger than 1. Items are presented one by one to be packed into suitable bins, where the information regarding bin types is known in advance. The cost of the algorithm is the total cost of used bins, and the goal is to minimize the cost. We show tight bounds on the absolute competitive ratio for the case of three types ( k=3 ), and tight bounds for the overall bound with an arbitrary number of types k, where those bounds are 2 and 4 for k=3 and for arbitrary k, respectively.
Operating rooms (ORs) are among the most resource-intensive and financially critical components of hospital infrastructure. Efficient surgery scheduling is of great importance to improve the utilization of ORs and reduce hospital expenditures. This paper addresses the surgery scheduling problem by integrating predictive analytics with prescriptive optimization. We first perform statistical learning to analyze the relationship between surgery duration and the volume of surgeries assigned to an operating room. Our analysis reveals an empirical predictive regularity: observed surgery durations tend to be lower in ORs assigned higher case volumes, up to a threshold of four cases. Based on these insights, we develop three optimization frameworks: a deterministic model, a stochastic programming model based on sample average approximation, and a robust optimization model. A simulation-based approach is proposed to calibrate the uncertainty budget in the robust optimization model. Numerical experiments demonstrate that models incorporating statistical learning consistently outperform those that do not. Furthermore, the robust optimization model provides superior or comparable performance to stochastic alternatives, offering a more reliable solution for high-uncertainty environments. By leveraging statistical learning, this study bridges the gap between surgical data analytics and resource optimization, providing a robust pathway for enhancing operating room efficiency.
A safe and efficient railway operation relies on a well-maintained system. The expenditure on railway track maintenance often contributes to approximately half of the maintenance expenditure; therefore, how to properly allocate the resources in railway track maintenance according to safety, comfort, and cost is an important task in the planning process. This study addresses the long-term track maintenance planning problem and aims to identify a set of maintenance tasks on rail networks for given horizons. In addition to the preventive maintenance strategy considered in the manual planning process, the predictive maintenance is taken into consideration to promote the performance of the railway track system. An optimization model is first established to describe the problem and a heuristic algorithm is then developed to enhance the solution efficiency based on the practical rules. Results from case studies demonstrate that the developed process can successfully identify appropriate maintenance tasks and assign them in the annual maintenance plan by minimizing the costs of expected failures (safety), irregularities (comfort), and maintenance. As a result, adopting this developed decision-support model can assist planners to identify the necessary maintenance activities and to also create long-term track maintenance plans efficiently.
The growing reliance on digital infrastructure has exposed financial institutions to increasingly dynamic cyber risks. As regulatory scrutiny and capital adequacy requirements intensify, there is rising demand for models that are both adaptive and interpretable. This paper proposes a hybrid framework that integrates regime-switching econometrics with explainable machine learning to model cyber attack severity in the financial sector. Using a curated subset of global cyber incidents affecting banks, insurers, and FinTech firms, we estimate two-regime Markov Switching Models (MSMs) with autoregressive components to identify latent regime dynamics in monthly incident intensity. Smoothed regime probabilities are then used to condition event-level classification models, allowing predictive relationships to vary across structural phases. Explainability is achieved through Shapley value analysis, enabling regime-specific interpretation of severity drivers. Empirical results identify a regime transition beginning in early 2023, characterized by increased persistence of high-severity activity and a reweighting of technical and actor-related predictors. The findings demonstrate that integrating econometric regime detection with interpretable machine learning provides a coherent and regime-sensitive framework for cyber risk modeling in regulated financial environments.
