Our research on data-driven capacity management focusses on developing prescriptive approaches for various important capacity planning problems such as staffing for logistics processes, call centers or public service offices. In these settings, we face different objectives such as minimizing capacity costs subject to service level constraints (like a maximum acceptable average waiting time) or minimizing costs for capacity, abandoning customers and waiting times. Our approaches directly learn optimal decision rules for a specific objective from collections of historical data. For this, we combine state-of-the-art machine learning techniques with stochastic optimization models from operations research. Hence, we overcome the traditional separated approach of first estimating a forecasting model of demand and subsequently solving a – not necessarily aligned - optimization problem given the demand predictions.
Lufthansa Technik Logistik Services: A maintenance service provider needs to staff its inbound operations for two shifts under demand uncertainty. Historical demand data of inbound deliveries of 120.000 individual shipments is available.
Main-Post Logistikgruppe: A logistics service provider needs to plan its staff capacity for mail sorting operations in a three-level-system with upgrading. Historical demand data over four years is available.
Call-Center staffing: A call center operator needs to plan its staff capacity for several shifts with the objective to minimize costs for capacity and abandoned calls. Historical data over one year from two call centers is available.
Public service office: A public service office needs to plan its staff capacity for two shifts per day under queueing constraints on expected waiting time. Historical data over two years is available
Prescriptive Analytics for a Queuing Model without Abandonment. in Available at SSRN 3516708 (2020).
Prescriptive Analytics for Flexible Capacity Management. in Available at SSRN 3387866 (2019).
Data-Driven Capacity Management with Machine Learning: A Novel Approach and a Case-Study for a Public Service Office. (2018). 105--115.