Computes (n-period ahead) dynamic forecasts of the signals and states for an estimated state space.
forecast computes the forecast for all observations in a specified sample. In some settings, you may instruct forecast to compare the forecasted data to actual data, and to compute summary statistics.
Syntax
sspace_name.forecast(options) keyword1 names1 [keyword2 names2] [keyword3 names3] ...
You should enter a type-keyword followed by a list of names for the target series or a wildcard expression, and if desired, additional type-keyword and target pairs. The following are valid keywords: @state, @statese, @signal, @signalse. The first two keywords instruct EViews to forecast the state series and the values of the state standard error series. The latter two keywords instruct EViews to forecast the signal series and the values of the signal standard error series.
If a list is used to identify the targets, the number of target series must match the number of names implied by the keyword. Note that wildcard expressions may not be used for forecasting signal variables that contain expressions. In addition, the “*” wildcard expression may not be used for forecasting signal variables since this would overwrite the original data.
Options
i=arg (default=”o”) | State initialization options: “o” (one-step), “e” (EViews computed), “u” (user-specified), “s” (smoothed). |
m=arg (default=“d”) | Basic forecasting method: “n” (n-step ahead forecasting), “s” (smoothed forecasting), “d” (dynamic forecasting. |
mprior = vector_name | Name of state initialization (use if option “i=u” is specified). |
n=arg (default=1) | Number of n-step forecast periods (only relevant if n-step forecasting is specified using the method option). |
vprior= sym_name | Name of state covariance initialization (use if option “i=u” is specified). |
prompt | Force the dialog to appear from within a program. |
p | Print view. |
Examples
The following command performs n-step forecasting of the signals and states from a sspace object:
ss1.forecast(m=n,n=4) @state * @signal y1f y2f
Here, we save the state forecasts in the names specified in the sspace object, and we save the two signal forecasts in the series Y1F and Y2F.
Cross-references
State space forecasting is described in
“State Space Models and the Kalman Filter”. For additional discussion of wildcards, see
Appendix A. “Wildcards”.