Presenter: Dr. Anindya Roy
In many applications, the underlying scientific theory may impose natural constraints on the parameters of commonly used models. While it is a good practice to maintain the constraints in any inferential procedure related to the parameters, it may turn out to be an extremely difficult proposition due to complexity of the constraints.
In such situations, it helps to reparameterize the problem in terms of parametric functions that
are free of constraints or at least have constraints that are more tractable. We will illustrate the advantages of making parameter transformation in the context of some messy constrained parameter problems.
Hosted by: here: Department of Biostatistics