Semiparametric Regression Smoothing of Non-linear Time Series (Q4255148)

From MaRDI portal

!

This is the item page for this Wikibase entity, intended for internal use and editing purposes. Please use the normal view instead:

scientific article; zbMATH DE number 1323323
Language Label Description Also known as
default for all languages
No label defined
    English
    Semiparametric Regression Smoothing of Non-linear Time Series
    scientific article; zbMATH DE number 1323323

      Statements

      Semiparametric Regression Smoothing of Non-linear Time Series (English)
      0 references
      10 August 1999
      0 references
      alpha-mixing
      0 references
      nonlinear time series
      0 references
      autoregressive model
      0 references
      semiparametric regression smoothing
      0 references
      Mackey-Glass system
      0 references
      0 references
      The author considers a special form of the following nonlinear time series models: \[ y_{t}=\bar g(y_{t-1},\ldots,,y_{t-r})+e_{t},\;y_{t}=\sum\limits_{i=1}^{p} \beta_{i}y_{t-c_{i}} +\sum\limits_{j=1}^{q} g_{j}(y_{t}-d_{j})+e_{t}, \quad t\geq\max(c_{p},d_{p})+1, \] where \(y_{t}\), \(e_{t}\in R\), \(g_{j}(\cdot)\) belong to a class of continuous functions from \(U\) (a subset of \(R\)) to \(R\), \(\{ g_{j}(\cdot)\colon 1\leq j\leq q\}\) are some unknown and nonlinear functions, the positive integers \(\{ (c_{i},d_{j})\colon 1\leq i\leq p, 1\leq j\leq q\}\) satisfy \(c_{i}\neq d_{j}\), \(1\leq c_1<\ldots<c_{p}\leq r\) and \(1\leq d_1<d_2\ldots<d_{q}\leq r\), and the \(e_{t}\)'s are i.i.d. random errors with \(Ee_{t}=0\) and \(Ee_{t}^2<\infty.\) For the particular case \(y_{t}=\beta y_{t-1}+ g(y_{t-2})+e_{t}\) of this model the author proposes an adaptive selection procedure for the number of summands in the series expansion of \(g(\cdot).\) An adaptive, data-driven estimator of \(\beta\) is constructed and two test statistics for testing \(H_0\colon \beta=0\) and \(H_{0}'\colon \sigma^2= \sigma_0^2\) (\(\sigma^2=Ee_{t}^2)\) in this model are established.
      0 references

      Identifiers