Decision-making with networks. Shift scheduling under uncertainty and smart, connected products

This thesis explores how networks can facilitate efficient decision-making in the face of complex structures and relationships. Specifically, networks are employed to represent shift generation rules for handling shift scheduling under uncertainty (Chapter 3 and Chapter 4), as well as the interdependencies of demand in the context of smart, connected products (Chapter 5). Chapter 3 introduces a novel approach by incorporating a state-expanded network into multi-activity shift scheduling under uncertainty. While providing a proof of concept, the computational results demonstrate the capability of the approach to handle shift flexibility. In cases where (partial) adjustments to shifts can be made after observing actual demand, this approach allows for flexible planning, leading to significant reductions in expected costs. The value of shift flexibility is intuitively derived from the efficient re-adjustment of contractually agreed working hours. Despite dealing with up to 10 activities and up to 100 scenarios, incorporating the rule-set introduced by Demassey et al. (2005), the approach effectively manages two shift types and easily accommodates restrictions related to them. The introduction of the novel and theoretically interesting mode with shift type commitment is a refinement of the well-known finding that considering uncertainty is valuable. Moreover, the study demonstrates that the combination of uncertainty and shift flexibility can be even more beneficial. Navigating a complex environment involves a trade-off with computational time. Chapter 4 introduces a novel approach, predict, tune, and optimize, which addresses this challenge by heuristically solving the problem. This approach tunes uncertain parameter predictions and problem-specific parameters to yield solutions for a deterministic optimization problem in a time-efficient manner, particularly for first-stage decisions. Unlike much of the prior work in this field, our approach is adept at handling uncertainty in the constraints. In Chapter 5, the focus shifts to the examination of smart, connected products and the associated inter-product network effects. We explore the idea that, in comparison to their traditional counterparts, smart products (i) generate additional consumer value and (ii) establish complementarities when used in conjunction with other smart products. The extent of complementarity among smart products may vary based on their nature, giving rise to an interproduct network. We incorporate these assumptions into a classical Cournot oligopoly model, where producers of smart products operate, and observe that equilibrium quantities are positively correlated with the Katz-Bonacich centrality of the inter-product network. In dissecting the equilibrium payoffs, we delve into the characterization of optimal pricing strategies for the supplier of the smart infrastructure.