dse (version 2020.2-1)

featherForecasts: Multiple Horizon-Step Ahead Forecasts

Description

Calculate multiple horizon-step ahead forecasts.

Usage

featherForecasts(obj, ...)
    # S3 method for TSestModel
featherForecasts(obj, data=NULL, ...)
    # S3 method for TSdata
featherForecasts(obj, model, ...)
    # S3 method for TSmodel
featherForecasts(obj, data, horizon=36,
             from.periods =NULL, ...)
    is.featherForecasts(obj)

Arguments

obj

an object of class TSmodel.

data

an object of class TSdata.

model

an object of class TSmodel.

from.periods

the starting points to use for forecasts.

horizon

the number of periods to forecast.

...

for a TSmodel additional arguments are passed to l()

Value

The result is a list of class featherForecasts with elements model (a TSestModel), data, from.periods, featherForecasts. The element featherForecasts is a list with length(from.periods) elements, each of which is a tframed matrix. There is a plot method for this class.

Details

Calculate multiple horizon-step ahead forecasts ie. use the samples indicated by from.periods to calculate forecasts for horizon periods. Thus, for example, the result of featherForecasts(model, data, from.periods=c(200,250,300)) would be forecasts for 1 through 36 steps ahead (the default), starting at the 200th,250th, and 300th point of outputData(data). This function assumes that inputData(data) (the exogenous variable) is as long as necessary for the most future forecast.

See Also

forecast, horizonForecasts

Examples

# NOT RUN {
data("egJofF.1dec93.data", package="dse")
model <- estVARXls(egJofF.1dec93.data)
pr <- featherForecasts(model, egJofF.1dec93.data)
tfplot(pr)
# }