Title :
Non-Gaussian state space modeling of time series
Author_Institution :
The Institute of Statistical Mathematics, Tokyo, Japan
Abstract :
A non Gaussian state space approach to the analysis of time series is shown. The model is expressed in general state space form which is expressed by a conditional distributions. General non-Gaussian filtering and smoothing formulae are shown and two numerical approximations to related distributions are used to realize these formulae. Significant merit of non Gaussian modeling is illustrated by some numerical examples.
Keywords :
Bayesian methods; Filtering algorithms; Gaussian distribution; Gaussian noise; Mathematical model; Mathematics; Nonlinear filters; Smoothing methods; State-space methods; Time series analysis;
Conference_Titel :
Decision and Control, 1987. 26th IEEE Conference on
Conference_Location :
Los Angeles, California, USA
DOI :
10.1109/CDC.1987.272759
