mixedSCORE: Mixed Membership Estimation
In zhoulijia/SCORE: Spectral Clustering On Ratios-of-Eigenvectors

Description Usage Arguments Details Value Author(s) References See Also Examples

View source: R/main.R

mixedSCORE(A,K,...) estimates the membership matrix in a network by the mixed SCORE (Spectral Clustering On Ratio-of-Eigenvector) algorithm.

1	mixedSCORE(A, K, L_theo = FALSE, mode = c("collapse", "each", "mutual") )

`A`	an igraph object that represents an undirected network.
`K`	an integer indicating the number of communities in the network
`L_theo`	logical. If `TRUE`, use the theoretical approach to estimate the tuning parameter L, which is the number of cluster to be used in the k-means step.
`mode`	an optional character constant that defines the conversion algorithm if the input igraph object A is directed

The input graph should be undirected. If the graph is directed, mixedSCORE will use as.undirected in the igraph package for conversion. By default, mixedSCORE uses the collapse mode. For details on the conversion algorithm, see as.undirected.

The leading eigenvector of the adjacency matrix of A should have strictly positive components. Otherwise, mixedSCORE will extract the giant component of the network, and positive components of the leading eigenvector are guaranteed by the Perron-Frobenius theorem.

mixedSCORE returns an object of class mSCORE, which is a list containing the following components:

`direct`	a character indicating the conversion algorithm if the graph is converted. If `NULL`, the input graph is undirected and does not need conversion.
`gc`	a list of character that contains the names of nodes that are not in the giant component, if the giant component is extracted from the network. If `NULL`, the giant component is not extracted.
`member`	the estimate of the n by K mixed membership matrix, where n is the number of nodes and K is the number of communities. The (i,j) entry corresponds to the probability that the i-th node belongs to the j-th community.
`l`	the optimal tuning parameter, which is an integer greater than K
`R`	the n by K-1 matrix whose columns represent the ratio of eigenvectors of the adjacency matrix.
`vertice`	the matrix whose columns are the estimated vertices in the vertice hunting step
`P`	the estimate of the K by K matrix P in the Degree Corrected Block model.
`B`	the estimate of the K by K matrix B in the Degree Corrected Block model.

Tracy Ke, Lijia Zhou and Qi Zhu.

Maintainer: Lijia Zhou <zlj@uchicago.edu>, Qi Zhu <qizhu@uchicago.edu>.

Jin, Ke and Luo (2017) "Estimating Network Membership By Simplex Vertices Hunting", Manuscript.

as.undirected, SCORE, and topicSCORE

# football contains the network of American football games 
# between Division IA colleges during regular season 
# Fall 2000, as compiled by M. Girvan and M. Newman.

# M. Girvan and M. E. J. Newman, Community structure in 
# social and biological networks, Proc. Natl. Acad. Sci. 
# USA 99, 7821-7826 (2002).

data(football)

fit = mixedSCORE(football,4) 
summary(fit)

fit$member