skewtFit-class: Class "skewtFit"

In twopiece: The family of two piece distributions

Description

Stores the Markov Chain Monte Carlo output from fitting a mixture of multivariate skew-t distributions. This is essentially a list with a few methods that were added for convenience (see below).

Details

Below is a description of what each element in the list contains:

- mu: List with G elements. Each contains Gibbs draws for the location parameter of each mixture component.

- Sigma: List with G elements. Each contains Gibbs draws for the scale matrix of each mixture component. Only diagonal and upper diagonal elements are stored, since the matrix is symmetric.

- alpha: List with G elements. Each contains Gibbs draws for the asymmetry parameters of each mixture component. Asymmetry parameters are in the epsilon-skew parameterization, see help(dtp3) for details.

- nu: Matrix with G columns, each row contains Gibbs draws for the degrees of freedom of each mixture component.

- probs: Matrix with G columns, each row contains Gibbs draws for the mixture weights.

- probcluster: Matrix with the estimated posterior probabilities that each observation belongs to each cluster (Rao-Blackwellized)

- cluster: if mixskewtGibbs was called with returnCluster=TRUE cluster contains the latent cluster allocations at each Gibbs iteration (iterations in rows, observations in columns). Else it contains NA.

- G: Assumed number of mixture components

- ttype: Value of argument ttype passed to mixskewtGibbs, 'independent' indicates that the iskew-t was used, 'dependent' the dskew-t.

Objects from the Class

Objects from this class are automatically created as the output of mixskewtGibbs.

Slots

.Data:

Single slot containing all the information. Since the class is built on a list one can access as usual its elements (mu, Sigma, alpha, nu, probs, probcluster, cluster, G, ttype). See details.

Methods

show

signature(object = "skewtFit"): Displays general information about the object.

"["

Selects an element of the list

"[["

Selects an element of the list

coef

signature(object = "skewtFit"): Computes the posterior mean of all model parameters

clusterprobs

signature(fit = "skewtFit"): Computes posterior cluster probabilities on a specified grid of value.

Author(s)

David Rossell

References

Rossell D., Steel M.F.J. Continuous non-Gassian mixtures. In Handbook of Cluster Analysis, CRC Press.

See Also

See also mixskewtGibbs.

Examples

showClass("skewtFit")

twopiece documentation built on May 2, 2019, 5:32 p.m.