forcavg |

Average different forecasts of a series.

Syntax

series.forcavg(options) forecast_data

You should specify the forecast data to be averaged by entering a list of objects as forecast_data. The list may be a list of series objects, a group object, a series naming pattern (such as “f*” to indicate all series starting with the letter “F”), or a list of equation objects.

If a list of equations is entered, EViews will automatically forecast from those equation objects over the forecast sample (the current workfile sample).

Options

wgt=”key” | Set the type of averaging to use. key can be “mean” (default), “trmean” (trimmed-mean), “med” (median), “ols” (least squares weights), “mse” (mean square error weights), “ranks”, (MSE ranks), “aic” (Smoothed AIC weights), or “sic” (BMA weights). “aic” and “sic” are only available if a list of equations is provided as the forecast_data. |

trim=num | Set the level of trimming for the Trimmed mean method. Num should be a number between 1 and 100. Only applicable if the “trmean” option is used. |

msepwr=int | Set the power to which the MSE values are raised in the MSE ranks method. Only applicable if the “mseranks” option is used. |

s | Use a static (rather than dynamic) forecast when computing the forecasts over the training sample. Only applicable if forecast_data is a list of equation objects. |

trainsmpl=arg | Specify the sample used for calculating the averaging weights. Only applicable if the “ols”, “mse”, “mseranks”, “aic” or “sic” options are used. |

name=arg | Set the name of the final averaged series. |

wgtname=arg | Save the weights into a vector in the workfile with the name wgtname. |

Example

The commands

wfopen elecdmd.wf1

elecdmd.forcavg(trainsmpl="2012M1 2012M12", wgt=mse) elecf_fe*

open the workfile elecdmd.wf1 and then perform forecast averaging using the actual series ELECDMD, and the forecast series specified by the naming pattern ELECF_FE*.

The averaging method MSE is used. A training sample of 2012M1 to 2012M12 is used to calculate the weights in the MSE and MSE Ranks methods.

See “Forecast Averaging” for additional discussion.

See also Series::forceval.