sgc: Run spectral clustering on a (possibly directed) (possibly...

In neurodata/graphstats: Graph Statistics

Description

It does

1. do a pass-to-rank for a weighted graph (PTR, no-op for an unweighted graph),

2. do a graph spectral embedding (ASE or LSE) with a diagonal augmentation,

3. do a dimension reduction (ZG) and merge left and right vectors (no-op for an undirected graph),

4. cluster vertices (GMM or Kmeans).

Usage

sgc(g, dmax = 2, elb = 1, abs = FALSE, lcc = TRUE, embed = "ASE",
  clustering = "GMM", Kmax = 9, weight = "raw", verbose = FALSE,
  doplot = FALSE)

Arguments

g	a graph in igraph format
dmax	maximum dimension for embedding
elb	an index for elbow
abs	a boolean to take abs on elbow finder or not
lcc	logical of whether to use the largest connected component or not.
embed	either ASE or LSE, spectral embedding method
clustering	either GMM or Kmeans, clustering method
Kmax	maximum number of clusters
weight	either ptr pr binary or raw to determine whether to perform pass-to-rank or not, default is raw
verbose	boolean to determine whether to display an intermediate fitting progress status of mclust or not, default is TRUE
doplot	boolean to determine whether to draw plots or not, default is TRUE

Value

g the largest connected component of the input graph

ase ASE or LSE output object

elb number of dimensions to which the graph was reduced.

mc clustering output object

Y labels for the clustering

Author(s)

Youngser Park [e-mail]

References

D.L. Sussman, M. Tang, D.E. Fishkind, and C.E. Priebe, A consistent adjacency spectral embedding for stochastic blockmodel graphs, Journal of the American Statistical Association, Vol. 107, No. 499, pp. 1119-1128, 2012.

Examples

library(igraph)
data(g)
E(g)$weight <- runif(ecount(g), 1, 5) # add random edge weights
Y <- gmmase(g, dmax=20, use.ptr=TRUE, embed="ASE", clustering="Kmeans")

neurodata/graphstats documentation built on May 14, 2019, 5:19 p.m.