dse: horizonForecasts – R documentation

Pricing

Become an expert in R — Interactive courses, Cheat Sheets, certificates and more!

Get Started for Free

Documentation

dse

horizonForecasts

Calculate forecasts at specified horizons

Description

Calculate forecasts at specified horizons.

Usage

is.horizonForecasts(obj)
    horizonForecasts(obj, ...)
    ## S3 method for class 'TSmodel'
horizonForecasts(obj, data, horizons=1:4,
	 discard.before=minimumStartupLag(obj), compiled=.DSEflags()$COMPILED, ...)
    ## S3 method for class 'TSestModel'
horizonForecasts(obj, data=NULL, ...)
    ## S3 method for class 'TSdata'
horizonForecasts(obj, model, ...)
    ## S3 method for class 'forecastCov'
horizonForecasts(obj,horizons=NULL, 
         discard.before=NULL, ...)

Arguments

`obj`	an object of class TSmodel, TSdata, or TSestModel.
`model`	an object of class TSmodel.
`data`	an object of class TSdata
`horizons`	a vector of integers indicating the horizon at which forecasts should be produced.
`discard.before`	period before which forecasts are not calculated.
`compiled`	if TRUE compiled code is called.
`...`	arguments passed to other methods.

Details

Calculate multiple 'horizon'-step ahead forecasts ie. calculate forecasts but return only those indicated by horizons. Thus, for example, the result of horizonForecasts(model, data horizons=c(1,5)) would be the one-step ahead and five step ahead forecasts.

Value

The result is a list of class horizonForecasts with elements model (a TSmodel), data, horizons, discard.before, and horizonForecasts. horizonForecasts is an array with three dimension: c(length(horizons), dim(model$data)). Projections are not calculated before discard.before or after the end of outputData(data). Each horizon is aligned so that horizonForecasts[h,t,] contains the forecast for the data point outputData(data)[t,] (from horizon[h] periods prior).

Examples

data("eg1.DSE.data.diff", package="dse")
model <- estVARXls(eg1.DSE.data.diff)
z <-  horizonForecasts(model, eg1.DSE.data.diff)

dse

Dynamic Systems Estimation (Time Series Package)

v2020.2-1

GPL-2

Authors

Paul Gilbert <pgilbert.ttv9z@ncf.ca>

Initial release