We introduce a model for a time series of continuous outcomes, thatcan be expressed as fully nonparametric regression or density regression on laggedterms. The model is based on a dependent Dirichlet process prior on a family ofrandom probability measures indexed by the lagged covariates. The approach isalso extended to sequences of binary responses. We discuss implementation andapplications of the models to a sequence of waiting times between eruptions ofthe Old Faithful Geyser, and to a dataset consisting of sequences of recurrenceindicators for tumors in the bladder of several patients. © 2013 International Society for Bayesian Analysis.
CITATION STYLE
Di Lucca, M. A., Guglielmi, A., Müller, P., & Quintana, F. A. (2013). A simple class of bayesian nonparametric autoregression models. Bayesian Analysis, 8(1), 63–88. https://doi.org/10.1214/13-BA803
Mendeley helps you to discover research relevant for your work.