|
Activity Number:
|
22
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Sunday, July 29, 2007 : 2:00 PM to 3:50 PM
|
|
Sponsor:
|
Section on Bayesian Statistical Science
|
| Abstract - #308546 |
|
Title:
|
Mixtures of State Space Models for Time Series with Structural Breaks
|
|
Author(s):
|
Ori Rosen*+ and Sally Wood and Robert Kohn
|
|
Companies:
|
University of Texas at El Paso and University of New South Wales and University of New South Wales
|
|
Address:
|
Department of Mathematical Sciences, El Paso, TX, 79968,
|
|
Keywords:
|
State-space Models ; Kalman Filter ; Mixtures-of-Experts ; MCMC
|
|
Abstract:
|
We consider mixtures of time series models which can be expressed in state space form. The mixing weights depend on time and on unknown parameters resulting in a flexible model that can accommodate varying local behavior of the time series. The number of components is treated as one of the model parameters. An example is a mixture of autoregressive models in which case both the common lag and the number of components are assumed unknown. We take a Bayesian approach and sample all the parameters from their posterior distribution. The methodology is illustrated with simulated data, as well as with real data.
|
