Conference Agenda

This paper discusses a decision support system (DSS) for the multi-level capacitated lot-sizing problem with linked lot sizes, backorders, and integrated shelf-life rules (MLCLSP-L-B-SL) applied to a pharmaceutical tablets manufacturing process. An exact mathematical formulation of the MLCLSP-L-B-SL is introduced. A Pareto analysis balancing costs, reliability of deliveries, and shelf-life requirements is outlined. Moreover, the developed DSS, its user interface, data management, and an optimization system are described. Outcomes of numerical experiments with real-world pharmaceutical tablets manufacturing processes are evaluated in terms of costs, service-level metrics, and expired inventories. Finally, planning rules and managerial insights are given for lot-sizing under shelf-life constraints.

Recently, specific e-commerce site management companies have taken on the responsibility of developing and operating e-commerce sites, and individual stores and companies that have previously conducted retail business only through brick-and-mortar stores will be able to utilize the developed platform and sell products on e-commerce sites by affiliating with the company. For organizations that sell products in this manner, it is necessary to consider which affiliated store to assign the requested order to after the e-commerce site company receives an order from a customer. There are several things to consider when assigning orders. Typical examples include whether the assigned affiliated store can fulfill all orders, whether the distance between the assigned affiliated store and the delivery destination is realistic. Currently, in some cases, operators manually perform part of the task of assigning orders to each affiliated store. To address the above problem, we developed an order assignment algorithm based on the proposed method using Voronoi tessellations, and thereby developed a system that automatically assigns orders to affiliated stores, replacing the current order assignment work currently performed manually by operators. We conducted numerical experiments using actual organizational data and verified how practical the results were by comparing to operator performance. As a result, the order assignment algorithm based on Voronoi tessellations showed superior results compared to operator performance.

The emerging technological innovations in AI, 5G, and information technologies have propagated new market opportunities for all industries, and the AI augmenting design and manufacturing of semiconductors have then been applied and spilled over to other related high-technology industries. This paper fills the gap in the industry econometrics literature that has not received much empirical attention in the presence of uncertainty and expectation mechanisms and the impacts on industrial dynamics. In markets with cyclical fluctuations and demand uncertainty, firms may have different dynamic decision rules facing upturns and downturns of industry cycles. We formulate a new dynamic framework with uncertainty and expectation mechanisms by integrating regime-switching industry cycles for cyclical fluctuation and demand uncertainty. Drawing on firm-level data of Taiwan's high-technology industries, we trace how Taiwan's high-technology companies upgraded their dynamic capabilities in facing asymmetric cyclical behavior and endogenous demand uncertainty. Explicitly incorporating the Markov regime-switching mechanism, we measure the firm’s dynamic adjustments when facing upturns and downturns of industry cycles. We also evaluate the firms’ dynamic decision rule and the resulting procyclical or counter-cyclical behavior in industry cycles. By unpacking the complex structure of industry cycles, the study extends the existing understanding of how some high-technology industries are more cyclical than others and attributes to capacity expansionary competition as a strategic competition. We conclude by highlighting the implications for research on the adjustment speed of essential inputs and the challenge of optimal forecasting of the expansionary and contractionary phases of the industry cycles.

This paper introduces several averages in dynamic team ratings in sports. In last year's presentation, we showed desirable properties in dynamic team ratings and proposed a simple model. However, this model had the disadvantage that the rating of a team in a given match depends only on the ratings of that team and the opposing team one match earlier, thus the ratings tend to fluctuate. Therefore, this study proposes "averages" equivalent to a moving average or an exponential smoother in normal time series. Specifically, we introduce daily and match windows and consider ratings by averages between them. Ratings based on these averages are applied to several team sports, and the results are compared. Finally, the final ratings obtained are compared with the results of the pre-season matches.