Skip to contents

Create MixAR objects

mixAR-methods.Rd

Generic function with methods for creating MixAR objects.

Usage

mixAR(template, coef, ..., filler = NA_real_)

Arguments

template

an object to be used as a template for the new object, typically inheriting from "MixAR". Alternatively, missing or a numeric vector specifying the order of the MixAR model, see Details.

coef

parameters for the new object a list with components "arcoef", "order", "prob", "shift", and "scale".

...

further arguments for methods.

filler

value for unspecified parameters, default is NA_real_.

Details

mixAR provides an alternative to the function new for specifying MixAR models.

If template is numeric vector, it is taken to specify the AR order of the model and the number of mixture components. A Gaussian MixAR model is created with parameters filled initially with NA's and then updated with values given by coef. coef does not need to have values for all parameters and may be missing altogether. If NA's are not suitable for initialisation, a suitable value can be specified with filler.

If template is a MixAR object, then the new object will have the class of template. The new object is set initially to a copy of template and then updated with parameters specified by coef (and maybe others for some methods).

In principle, the numeric parameters are vectors of length the number of components of the MixAR model. For convenience, single values are replicated to the number of components. For this to work, at least one component must be specified completely, for example the order. It is an error for the parameters to imply conflicting number of components.

Methods

signature(template = "ANY")

signature(template = "MixAR")

See also

class "MixARGaussian", class "MixARgen"

Examples

mixAR(coef = list(prob = c(.5,.5), scale = c(1,2), 
                  arcoef = list(.5, 1.1), shift = c(0,0), order = c(1,1)))
#> An object of class "MixARGaussian"
#> Number of components: 2 
#>        prob shift scale order ar_1
#> Comp_1 0.5    0     1     1   0.5 
#> Comp_2 0.5    0     2     1   1.1 
#> 
#> Distributions of the error components:
#> 	standard Gaussian
#> 

mixAR(template = c(1,1))
#> An object of class "MixARGaussian"
#> Number of components: 2 
#>        prob shift scale order ar_1
#> Comp_1  NA   NA    NA     1    NA 
#> Comp_2  NA   NA    NA     1    NA 
#> 
#> Distributions of the error components:
#> 	standard Gaussian
#> 
mixAR(coef = list(order = c(1,1))) # same
#> An object of class "MixARGaussian"
#> Number of components: 2 
#>        prob shift scale order ar_1
#> Comp_1  NA   NA    NA     1    NA 
#> Comp_2  NA   NA    NA     1    NA 
#> 
#> Distributions of the error components:
#> 	standard Gaussian
#> 

m2 <- new("MixARGaussian", order = c(3, 2, 1),
          arcoef = matrix(c(1:3, c(1:2, 0), c(1, 0, 0)), nrow = 3, byrow = TRUE))
m2a <- mixAR(m2, list(prob = c(0.5, 0.25, 0.25)))
show_diff(m2, m2a)
#>        prob shift scale order ar_1 ar_2 ar_3
#> Comp_1  NA    0     0     3     0    0    0 
#> Comp_2  NA    0     0     2     0    0      
#> Comp_3  NA    0     0     1     0