Seminars 2007 — Abstracts
Friday, July 27
Speaker:
Geoff Robinson,
CSIRO
Title:
Models with continuous-time Brownian motion as well as white noise
Abstract: This talk is primarily concerned to answer versions of the question "What's been happening?" given sequential data from variable processes, particularly when the data are not equally-spaced in time. It concentrates on models with stochastic parts compromising Brownian motion as well as white noise, and presents some simple Time Series methodology in a non-standard way. The concepts and techniques relied upon most heavily are mixed models, Kalman filtering and variograms. The approach adopted leads to clear opinions about acceptance control, Statistical Process Control, interpreting the EWMA, single source of error models for discrete-time data, the local linear trend model, and the relative merits of State space models and ARIMA models.
