Current developments in time series modelling

Abstract

In this paper a survey is given of some current developments in time series analysis. After a brief discussion of linear models, ARMA models and state–space reprentations, the discussion concentrates on a fairly general class of non-linear models called ‘state-dependent models’ and the applications of these to the problem of identifying general forms of non-linear structure. The effectiveness of the estimation technique is studied by means of numerical examples on simulated data. Towards the end of the paper it is indicated how these ideas may be applied to the study of systems governed by non-linear differential equations.