Linear Parameterization Of Dynamic Models for Time Series Analysis and Forecasting
by
Akram M. Chaudhry
University of Bahrain, Sukhir, Bahrain

While constructing linear dynamic system models for analysis and forecasting of discrete time series there, always, exists a problem of parameterization of models. This problem becomes more severe when models are required to be constructed, in a parsimonious manner, to filter the pollution of colored noise processes, such as, auto-regressive and moving average type processes.

To solve this problem, a procedure for parameterization of models is discussed. This, simple and straight forward procedure, is expected to assist the model builders in the construction of dynamic models, parameterize or/and re-parameterize them in an appropriate manner. Equivalently, it is also expected to help practitioners in understanding the nature and behavior of parameters and yield optimum results, in both the canonical and the diagonal form of models.

Date received: November 14, 2001

Copyright © 2001 by the author(s). The author(s) of this document and the organizers of the conference have granted their consent to include this abstract in Atlas Conferences Inc. Document # caid-93.