Title: Dependent random measures indexed by a functional covariate
Authors: Emmanuel Bernieri - University of Edinburgh (United Kingdom) [presenting]
Miguel de Carvalho - CEAUL (Centro de Estatistica e Aplicacoes), Universidade de Lisboa (Portugal)
Abstract: The analysis of functional data is explored in a nonparametric Bayesian context. Specifically, we devise priors in the space of all conditional distributions, for the setting where the interest is on conditioning on a sophisticated object such as a random function. The proposed model can be regarded as an infinite mixture of functional linear regression models. A specific version of the proposed model is explored in detail, which consists of a Dependent Dirichlet Process (DDP) whose regression functions include inner products between a functional covariate and coefficient function. We illustrate the proposed methods using simulated and real data, and evaluate the accuracy of the methods through a simulation study.