Highlights A review on operations research (OR) models and methods for safety stock determination is conducted. No work has yet systematized research focusing on the safety stock determination problem. Articles are classified and discussed regarding the modeling approach, industrial application, solution technique and main performance criteria used. Research opportunities, promising research directions and trends are identified.

The constant advancements in Information Technology have been the main driver of the Big Data concept’s success. With it, new concepts such as Industry 4.0 and Logistics 4.0 are arising. Due to the increase in data volume, velocity, and variety, organizations are now looking to their data analytics infrastructures and searching for approaches to improve their decision-making capabilities, in order to enhance their results using new approaches such as Big Data and Machine Learning. The implementation of a Big Data Warehouse can be the first step to improve the organizations’ data analysis infrastructure and start retrieving value from the usage of Big Data technologies. Moving to Big Data technologies can provide several opportunities for organizations, such as the capability of analyzing an enormous quantity of data from different data sources in an efficient way. However, at the same time, different challenges can arise, including data quality, data management, and lack of knowledge within the organization, among others. In this work, we propose an approach that can be adopted in the logistics department of any organization in order to promote the Logistics 4.0 movement, while highlighting the main challenges and opportunities associated with the development and implementation of a Big Data Warehouse in a real demonstration case at a multinational automotive organization.

Demand forecasting works as a basis for operating, business and production planning decisions in many supply chain contexts. Yet, how to accurately predict the manufacturer's demand for components in the presence of end-customer demand uncertainty remains poorly understood. Assigning the proper order quantities of components to suppliers thus becomes a nontrivial task, with a significant impact on planning, capacity and inventoryrelated costs. This paper introduces a multivariate approach to predict manufacturer's demand for components throughout multiple forecast horizons using different leading indicators of demand shifts. We compare the autoregressive integrated moving average model with exogenous inputs (ARIMAX) with Machine Learning (ML) models. Using a real case study, we empirically evaluate the forecasting and supply chain performance of the multivariate regression models over the component's life-cycle. The experiments show that the proposed multivariate approach provides superior forecasting and inventory performance compared with traditional univariate benchmarks. Moreover, it reveals applicable throughout the component's life-cycle, not just to a single stage. Particularly, we found that demand signals at the beginning of the life-cycle are predicted better by the ARIMAX model, but it is outperformed by ML-based models in later life-cycle stages.

The complexity of optimal control problems requires the use of numerical methods to compute control and optimal state trajectories for a dynamical system, aiming to optimize a particular performance index. Considering a real viral advertisement, this article compares the dynamics of a viral marketing epidemic model with optimal control under different cost scenarios and from two perspectives: using numerical methods based on the Pontryagin's Maximum Principle (indirect methods) and methods that treat the optimal control problem as a nonlinear constrained optimization problem (direct methods). Based on the trade-off between the maximization of information spreading and the minimization of the costs associated with it, an optimal control problem is formulated and studied. The existence and uniqueness of the solution are proved. Our results show not only that the cost of implementing control policies is a crucial parameter for the spreading of marketing messages, but also that low investment costs in control strategies fulfill the proposed trade-off without compromising the financial capacity of a company.

At a time where available data is rapidly increasing in both volume and variety, descriptive Data Mining (DM) can be an important tool to support meaningful decision-making processes in dynamic Supply Chain (SC) contexts. Up until now, however, scarce attention has been given to the application of DM techniques in the field of inventory management. Here, we take advantage of descriptive DM to detect and grasp important patterns among several features that coexist in a real-world automotive electronics SC. Concretely, Principal Component Analysis (PCA) is employed to analyze and understand the interrelations between ten quantitative and dependent variables in a multi-item/multi-supplier environment. Afterwards, the principal component scores are characterized via a K-means clustering, allowing us to classify the samples into four clusters and to derive di↵erent profiles for the multiple inventory items. This work provides evidence that descriptive DM contributes to find interesting feature-patterns, resulting in the identification of important risk profiles that may e↵ectively leverage inventory management for superior performance.

Purposing to lessen malware propagation, this paper proposes optimal control measures for a susceptible-carrier-infectious-recovered-susceptible (SCIRS) epidemiological model formed by a system of ordinary differential equations. By taking advantage of real-world data related to the number of reported cybercrimes in Japan from 2012 to 2017, an optimal control problem is formulated to minimize the number of infected devices in a cost-effective way. The existence and uniqueness of the results related to the optimality system are proved. Overall, numerical simulations show the usefulness of the proposed control strategies in reducing the spread of malware infections.

The process of diffusing viral marketing campaigns through social networks can be modeled under concepts of mathematical epidemiology. Based on a Susceptible-Infected-Recovered (SIR) epidemiological model, the benefits of optimal control theory on the diffusion of a real viral advertisement are studied. Two optimal control strategies that could help marketers to maximize the spread of information and minimize the costs associated to it in optimal time windows are analyzed and compared. The uniqueness of optimality system is proved. Numerical simulations show that high investment costs in publicity strategies do not imply high overall levels of information diffusion. This paper contributes to the current literature by studying a viral marketing campaign using real numerical data.

No abstract