This project deals with the development of methods for assisting a decision analyst in establishing a model for some problem domain. For example, when modeling a decision problem it can be advantageous for the decision analyst to be able to identify and focus on local parts of the problem rather than working with the decision problem as a whole. One way of achieving this is to identify the set of variables which is relevant for a particular decision variable, i.e., the set of variables that may potentially influence that decision. This analysis could then be supplemented with an analysis of how sensitive the model is to variations in the utility and probability parameters; if the model is particular sensitive w.r.t. a certain parameter, then the analyst can focus on this parameter during the elicitation process.
|Periode||19/05/10 → …|