varmaPredict#

Purpose#

Calculates forecasts from a VARMAX model.

Format#

f = varmaPredict(vmo, y, x, t)#

Parameters:

vmo (struct) – An instance of avarmamtOut structure returned by a call to a VARMAX or ECM procedure.
y (TxL matrix) – the variables to be forecast.
x (txK matrix) – x covering only the forecast horizon, in the order or the scalar zero if there are no x variables.
t (scalar) – the number of periods to forecast.

Returns:

f (tx(L+1) matrix) – Column one contains the period forecast. The remaining columns contain the forecast values.

Example#

new;
cls;
library tsmt;

// Load data
// Create file name with full path
fname = getGAUSSHome() $+ "pkgs/tsmt/examples/mink.csv";

// Load two variables from dataset
y = loadd(fname, "LogMink + LogMusk");

// Difference the data
y = vmdiffmt(y, 1);

// Declare 'vout' to be a varmamtOut structure
struct varmamtOut vout;

// Estimate the parameters of the VAR(2) model
vout = varmaFit(y, 2);

// Predict model
f = varmaPredict(vout, y, 0, 25);

Remarks#

The varmaFit and ecmFit procedures estimate centered models and do not return intercepts. However, varmaPredict allows intercepts, so that it might be used with the results of other estimation procedures.

Library#

tsmt

Source#

varmamt.src