Abstract: In this chapter we present nonparametric methods and available quantlets for nonlinear modelling of univariate time series. A general nonlinear time series model for an univariate stochastic process Y t t=1 T is given by the heteroskedastic nonlinear autoregressive (NAR) process $${Y_t} = f\left( {{Y_{t - i1}},{Y_{t - i2}}, \ldots ,{Y_{t - im}}} \right) + \sigma \left( {{Y_{t - i1}},{Y_{t - i2}}, \ldots ,{Y_{t - im}}} \right){\xi _t},$$ (16.1) where ξ t denotes an i.i.d. noise with zero mean and unit variance and ƒ(·) and σ(·) denote the c...
(read more)
Topics: 
Applied mathematics
Statistics
Combinatorics