cluster_spectral {HyperG}

Spectral Graph Clustering

Description

Use spectral embedding to embed a graph into a lower dimension, then cluster the points using model based clustering. This results in a clustering of the vertices.

Usage

cluster_spectral(g, verbose = FALSE, adjust.diag = FALSE, laplacian = FALSE, 
   normalize = FALSE, scale.by.values = FALSE, vectors = "u", d = 12, ...)

Arguments

g

a graph.

verbose

logical. Whether to print to the screen as it goes.

adjust.diag

logical. Whether to set the diagonal of the adjacency matrix to degree/(n-1).

laplacian

logical. Whether to use the Laplacian rather than the adjacency matrix.

normalize

logical. Whether to normalize the matrix by D^1/2.

scale.by.values

Whether to scale the embedding vectors by the eigen vectors.

vectors

character. "u" or "v" or "uv". The latter is only appropriate for directed graphs.

d

embedding dimension.

...

arguments passed to Mclust.

Details

This first embeds the vertices into a d-dimensional space, using the adjacency matrix or the Laplacian. See ase for more information. It then applies Mclust to the resultant points to cluster.

Value

An object of class "Mclust".

Author(s)

David J. Marchette [e-mail]

References

Fraley C. and Raftery A. E. (2002) Model-based clustering, discriminant analysis and density estimation, _Journal of the American Statistical Association_, 97/458, pp. 611-631.

See Also

ase.

Examples

	P <- rbind(c(.2,.05),c(.05,.1))
	ns <- rep(50,2)
	set.seed(451)
	g <- sample_sbm(sum(ns),P,ns)
	cluster_spectral(g)