Univariate nonlinear time series models

In this paper developments in the analysis of univariate nonlinear time series are considered. First a number of commonly used nonlinear models are presented. The next section is devoted to methods of testing linearity, which is an important part of nonlinear model building. Techniques of modelling nonlinear series within a predetermined family of models are discussed thereafter. Forecasting with nonlinear models also has its own section. A brief set of final remarks closes the chapter. -- Hidden Markov model ; linearity test ; neural network ; nonlinear model building ; threshold autoregressive model ; smooth transition autoregressive model