Using a Bayesian Network in the ProdFLOWTM Approach

ProdFLOWTM1 is a new approach for the productivity analysis and management of research & development organisations created by the research department of the Siemens AG. Its core are organisation-specific models based on the respective substantial levers of productivity. Levers that are both influenceable and measurable are compiled together with the experts of the organisation. This paper proposes to use Bayesian networks for building such models. It is shown how the networks are structured, how they are parameterised and used to analyse different improvement scenarios. Experiences from a case study suggest that Bayesian networks are a suitable technique for the organisation-specific models in ProdFLOWTM.