The Dependent Dirichlet Process and Related Models

Standard regression approaches assume that some finite number of the response distribution characteristics, such as location and scale, change a (parametric or nonparametric) function predictors. However, it is not always appropriate to location/scale representation, where error has unchanging shape over predictor space. In fact, often happens in applied research responses under study changes with predictors ways cannot be reasonably represented by dimensional functional form. This can seriously affect answers scientific questions interest, therefore more general are indeed needed. gives rise fully nonparametric models. We review main Bayesian have been employed define probability models complete may vary flexibly focus on developments based modifications Dirichlet process, historically termed dependent processes, extensions proposed tackle this problem using approaches.