Probability Management: Revisiting an Old Approach to Business Modelling Under Uncertainty
Over the past decade, optimization methodology has made significant contributions to financial engineering (stochastic programming) and to operations management (integer programming). In contrast another very important area in business – business planning – has been largely unaffected by our methods. I will first step back and highlight some of the prevalent pitfalls in business planning that permeate many if not all industries and offer significant impact opportunity for stochastic modelling and optimization. A key obstacle to a successful use of mathematical models in business decision making is the tendency to build big central models, which eventually collapse under their own weight. What is needed are agile networks of small local models that can talk to one-another. I will illustrate that a simple approach to modelling uncertainty may help us to build such decentralised networks of models and will comment on some preliminary industry experience. This talk is based on joint work with Sam Savage (Stanford) and Daniel Zweidler (Shell).
Stefan Scholtes, Cambridge University, United Kingdom
