Topics in Functional Time Series

• Activity Number: 639
• Type: Invited
• Date/Time: Thursday, August 2, 2012 : 10:30 AM to 12:20 PM
• Sponsor: IMS
• Author(s): Alexander Aue*+
• Companies: University of California at Davis
• Keywords: autoregressive process ; break points ; functional data ; functional linear model ; model selection

Abstract

Modern applications of statistics often involve observations that are conveniently viewed as curves or functional data. In the recent past, a rich theory has emerged that is widely used in many areas of science with a statistical component. Motivated by applications in finance, geophysics and civil engineering, we discuss in this talk several new lines of research for functional time series data which include the handling of nonstationarity through a segmentation approach, the inclusion of exogenous covariates into the functional autoregressive process framework, as well as model selection and order selection in functional linear and functional autoregressive models, respectively. For all cases, supporting large sample theory will be given. Simulation evidence and applications, for example the modeling of the average daily flow volume for the Mokelumne River in the central Sierra Nevada, are also presented.