Title: Introduction to Lancaster University and STOR-i CDT | Using the recourse space for scenario reduction in stochastic programming

Speaker: Dr Anna-Lena Sachs OPTIMA PI; Hamish STOR-i PhD student

Summary: An introduction to the programs at STOR-i at Lancaster University, followed by a research-focused presentation by Hamish Thorburn. STOR-i Centre for Doctoral Training is a joint venture between the departments of Mathematics and Statistics and Management Science. It offers students a four-year PhD programme in Statistics and Operational Research (STOR) developed and delivered with industrial partners. OPTIMA students who are interested in undertaking a paid exchange at Lancaster are encouraged to attend.

Many important decisions in practice such as staff allocation, facility location or network design need to be made before all information is known. These problems are often solved using stochastic programming with a given scenario set to take uncertainty into account. Finding representative scenario sets for stochastic programming is key for finding good quality solutions within a reasonable time. We are interested in scenario reduction methods – methods which reduce a large scenario set to a smaller one. Often this reduction is done based on how similar the scenarios are. However, this can lead to suboptimal results, as similarity between scenarios does not guarantee similarity in how they respond to different solutions. We instead use recourse similarity of scenarios when using scenario reduction, to select more representative scenarios. We do this by transforming scenarios to the “recourse space” where distances between scenarios represent distances in recourse similarity. Here, we present our work transforming solutions to the recourse space to improve scenario reduction methods. We also apply this method to several practical problems and show how it can improve results.

Biography: Dr Anna-Lena Sachs is a Senior Lecturer in Predictive Analytics at Lancaster University. Her research focuses on inventory management including the intersection to forecasting and behavioural operations management. With her research, she has informed the practices at companies in different industries such as retailing, transportation and health care. She developed several quantitative models allowing decision-makers making better inventory decisions. In field experiments and conducts laboratory she analyses what type of decision support is useful in practice and how human decision makers actually behave.

Hamish Thorburn is an Australian PhD student at Lancaster University. Growing up in Brisbane, he first attended university at the University of Queensland, before moving to the UK for postgraduate study in September 2019.

MEETING ID: 873 1557 5255; PASSWORD: 778635

WED 14 June 16:00-17:00 (AEST, Melbourne Time) – online only