fit_mixVAR-methods {mixAR}

Fit mixture vector autoregressive models

Description

Estimate a MixVAR model for a multivariate time series. This is a generic function. The methods defined in package MixAR are described here.

Usage

fit_mixVAR(x, model, fix, ...)

Arguments

x

a multivariate time series (currently a numeric matrix).

model

model, object inheriting from MixVAR class.

fix

if TRUE, fix the shift parameters.

...

additional arguments for the methods (not currently used).

Details

model specifies the model to fit. If model inherits from "MixVAR", it is used as a template. Estimation is done via EM-Algorithm, using the function mixVARfit.

Currently the default method for fit_mixAR just throws error, since there seems no suitable default task to do.

Value

a MixVAR model.

Methods

signature(x = "ANY", model = "MixVAR")

signature(x = "ANY", model = "ANY")

See Also

mixVARfit

Examples

AR <- list()
AR[[1]] <- array(c(0.5, -0.3, -0.6, 0, 0, 0.5, 0.4, 0.5, -0.3), dim = c(3, 3, 1))
AR[[2]] <- array(c(-0.5, 0.3, 0, 1, 0, -0.5, -0.4, -0.2, 0.5), dim = c(3, 3, 1))

prob <- c(0.75, 0.25)
shift <- cbind(c(0, 0, 0), c(0, 0, 0))

Sigma1 <- cbind(c(1, 0.5, -0.4), c(0.5, 2, 0.8), c(-0.4, 0.8, 4))
Sigma2 <- cbind(c(1, 0.2, 0), c(0.2,  2, -0.15), c(0, -0.15, 4))
Sigma <- array(c(Sigma1, Sigma2), dim = c(3, 3, 2))

m <- new("MixVARGaussian", prob = prob, vcov = Sigma, arcoef = AR, shift = shift)

set.seed(1234)
y <- mixVAR_sim(m, n = 100, init = matrix(0, ncol = 3), nskip = 50, flag = FALSE)

fit_mixVAR(y, m, tol = 1e-3)
mixVARfit(y, m, tol = 1e-3)