Doctoral Thesis
Modelling joint autoregressive moving average processes
Doctor of Philosophy (PhD), Murdoch University
2017
Abstract
This thesis explores Joint Autoregressive Moving-Average (JARMA) models for independent replicated univariate time series with common ARMA coefficients whose innovations variances are either in common, unique to each series or vary with the series mean. The constraint of a common variance is also applied to vector ARMA processes. Interleaving is shown to represent replicated series with a common variance as one series from the same process. The time and frequency domain properties of interleaved replicated stationary and invertible processes are established. As an aid to identification, hypothesis tests for comparing series are reviewed and several new tests are presented and explored along with a graphical method for identification. Unconditional maximum likelihood estimates of the parameters of various JARMA processes are derived using the methods of joint likelihood and interleaving. The properties of the estimators are examined using simulation and asymptotics. Finally JARMA models are fitted to over 60 years of daily univariate and bivariate temperature data to estimate differences in level due to location and climate change.
Details
- Title
- Modelling joint autoregressive moving average processes
- Authors/Creators
- Ross Bowden (Author/Creator)
- Contributors
- Brenton Clarke (Supervisor)Nicola Armstrong (Supervisor)
- Awarding Institution
- Murdoch University; Doctor of Philosophy (PhD)
- Identifiers
- 991005541810107891
- Murdoch Affiliation
- School of Engineering and Information Technology
- Language
- English
- Resource Type
- Doctoral Thesis
Metrics
205 File views/ downloads
145 Record Views