Online Program Home
My Program

Abstract Details

Activity Number: 488 - Nonstationary and Anisotropic Spatial Processes
Type: Contributed
Date/Time: Wednesday, August 1, 2018 : 8:30 AM to 10:20 AM
Sponsor: Section on Statistics and the Environment
Abstract #329726 Presentation
Title: Exploring Departures from Stationarity Using Locally Stationary Time Series
Author(s): Shreyan Ganguly* and Peter Craigmile
Companies: The Ohio State University and The Ohio State University
Keywords: Locally stationary processes; uncertainty quantification; parameter estimation; basis functions; climate

The assumption of stationarity is appealing in classical time series analysis because of the ease of estimation and forecasting. However, stationarity is an idealization which, in practice, can at best hold as an approximation, but for many time series may be an unrealistic assumption. In this talk, we define a class of locally stationary processes which can lead to improved forecasts and more accurate uncertainty quantification. This class of processes assumes the model parameters to be time-varying and parameterizes them in terms of a transformation of basis functions that ensures that the processes are locally stationary. We investigate methods and theory for parameter estimation in this class of models, and propose tests of stationarity that have the potential to allow us to examine certain departures from stationarity in time series data. We assess our methods using simulation studies and apply these techniques to the analysis of climate time series.

Authors who are presenting talks have a * after their name.

Back to the full JSM 2018 program