stl |

Seasonally adjust series using the STL decomposition method.

Unlike other seasonal adjustment methods used by EViews, this procedure works on any time frequency.

Syntax:

series.stl(options) seas_name[trend_name]

You should follow the stl keyword with a name for the seasonally adjusted series. Optionally, you may also provide a name for the output trend series.

Options

Periodicity=arg | Specify the periodicity. Use āwā to expand weekly data to 53 weeks and ādā to expand daily data to 366 (in a 7 day week workfile) or 261 (in a 5 day week workfile) days. Default is the number of periods per year (expanded for weekly and daily). |

Sp=integer | Specify the seasonal polynomial degree. Default is 0. |

tp=integer | Specify the trend polynomial degree. Default is 1. |

fp=integer | Specify the filter polynomial degree. Default is 1. |

Sl=integer | Specify the length of the seasonal smoothing window (odd integers only). Default is 35. |

Tl=integer | Specify the length of the trend smoothing window (odd integers only). Default is based upon the seasonal smoothing window length. |

Fl=integer | Specify the length of the filter smoothing window (odd integers only). Default is based upon the data frequency. |

Inits=integer | Specify number of inner iterations. Default is 5. |

Outits=integer | Specify the number of outer iterations. Default is 15. |

Estsmpl=arg | Set the estimation sample. |

Forclen=integer | Set the length of the forecast. |

Seasdiagnostic | Display seasonality diagnostics graph. |

Examples

Co2.stl co2_sa c02_trend

performs STL decomposition on the series C02, saving the adjusted data in the series C02_SA and the trend in C02_TREND.

show Co2.stl(sl=20, outits=20, seasdiagnostic)

performs the same decomposition, but with a seasonal smoothing window of 20, using 20 iterations of the outer loop, and displays the seasonal diagnostics graphs.