Online Program Home
My Program

Abstract Details

Activity Number: 84
Type: Contributed
Date/Time: Sunday, July 31, 2016 : 4:00 PM to 5:50 PM
Sponsor: Section on Bayesian Statistical Science
Abstract #320588
Title: A Bayesian Approach to Measuring the Nonstationarity of a Time Series
Author(s): Sourav Das* and Guy Nason
Companies: and University of Bristol
Keywords: Bayesian DLM ; non-parametric regression ; roughness penalty

Since the 1960's non-stationary time series have been investigated extensively. Methodology and theory have evolved rapidly since Dahlhaus' construction of locally stationary processes in the 1990's. However much of the theory in above constructions rely on assumptions of smoothness on the time varying transfer function. However when modelling real data, tools for assessing such regularity conditions are yet to be developed. We propose a methodology for measuring the degree of non-stationarity in a time series. We use principles of non-parametric regression to measure the roughness of posterior distribution of the underlying signal and show its association with stationarity. The method is tested on simulated time-varying autoregressive processes. We also demonstrate potential applications of this index in finding solutions to pertinent questions in Earth Sciences.

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

Back to the full JSM 2016 program

Copyright © American Statistical Association