Files
Abstract
Increasing complexity in research topics poses problems to the communication of insights with stakeholders. A method to overcome this obstacle is to interactively execute modelling techniques. As the underlying simulation models are usually too complex for on spot modelling, new approaches are needed. One approach is to apply meta models based on Bayesian models that capture the behaviour of more complex models within its conditional probability tables. This paper presents the methodology to set up the necessary tables and to include third party information seamlessly. To test the resulting model, test persons were exposed a graphical user interface for the Bayesian model with the task to fix certain agronomic parameters in a situation of climate change. Their optimizing performance and feedback on the modelling experience was evaluated. Results show that the usability and comprehensibility of the model was high. Whether the models will be trusted by the stakeholders will depend on the underlying data that was fed into the model.