make_reduced {concorR}

Run CONCOR and output reduced adjacency matrices

Description

Takes a list of adjacency matrices, partitions using concor, and returns a list of reduced adjacency matrices and their corresponding cutoff densities. Each reduced matrix corresponds to one input relation. The sna package must be installed, but does not need to be attached.

Usage

make_reduced(adj_list, nsplit = 1)

Arguments

adj_list

A list of adjacency matrices, each representing a different relation for the same nodes.

nsplit

The total number of splits CONCOR will (attempt to) perform.

Details

A reduced network represents each identified position in the network as a single node. Links (or self-links) exist if the density of ties in that block is greater than a threshold density. In this implementation, the density of the whole network is used as the threshold for each block.

In the list of input matrices adj_list, each should correspond to a different relation for the same nodes. Each adjacency matrix is partitioned with the CONCOR algorithm, continuing for nsplit divisions. After the threshold density is applied, each entry in the reduced matrix has values of 0 or 1.

Value

reduced_mat

A list of reduced matrices, one for each input matrix.

dens

A vector of the cut-off densities used (equal to the edge density of each entry in adj_list).

References

S. Wasserman and K. Faust, Social Network Analysis: Methods and Applications (Cambridge University Press, 1994).

See Also

concor, make_blk

Examples

g1 <- matrix(c(0, 0, 0, 1, 1, 0, 1, 0, 1, 1, 0, 1, 1, 0, 1, 0), ncol = 4)
rownames(g1) <- c("a", "b", "c", "d")
colnames(g1) <- c("a", "b", "c", "d")

g2 <- matrix(c(0, 1, 0, 0, 0, 0, 1, 0, 0, 0, 1, 1, 1, 1, 0, 0), ncol = 4)
rownames(g2) <- c("a", "b", "c", "d")
colnames(g2) <- c("a", "b", "c", "d")

make_reduced(list(g1, g2), nsplit = 1)