Count Network Statistics — count

This function is similar to what ergm::summary_formula does, but it provides a fast wrapper suited for matrix class objects (see benchmark in the examples).

count_stats(X, ...)

AVAILABLE_STATS()

# S3 method for formula
count_stats(X, ...)

# S3 method for list
count_stats(X, terms, attrs = NULL, ...)

Arguments

X	List of square matrices. (networks)
...	Passed to the method.
terms	Character vector with the names of the statistics to calculate. Currently, the only available statistics are: 'mutual', 'edges', 'ttriad', 'ctriad', 'ctriple', 'nodeicov', 'nodeocov', 'nodematch', 'triangle', 'balance', 't300', 't102', 'absdiff', 'idegree1.5', 'odegree1.5', 'ostar1', 'ostar2', 'ostar3', 'ostar4', 'istar1', 'istar2', 'istar3', 'istar4 '.
attrs	A list of vectors. This is used when `term` has a nodal attribute such as `nodeicov(attrname="")`.

Value

A matrix of size length(X) * length(terms) with the corresponding counts of statistics.

Examples

# DGP 
set.seed(123199)
x <- rbernoulli(rep(5, 10))
ans0 <- count_stats(x, c("mutual", "edges"))

# Calculating using summary_formula
ans1 <- lapply(x, function(i) {
  ergm::summary_formula(i ~ mutual + edges)
})

ans1 <- do.call(rbind, ans1)

# Comparing
all.equal(unname(ans0), unname(ans1))
#> [1] TRUE

# count_stats is vectorized (and so faster)
bm <- benchmarkito(
  count_stats = count_stats(x, c("mutual", "edges")),
  lapply      = lapply(x, function(i) {
  ergm::summary_formula(i ~ mutual + edges)
}), times = 50
)
#> Error in lapply(x, function(i) {    ergm::summary_formula(i ~ mutual + edges)}): object 'x' not found

plot(bm)
#> Error in plot(bm): object 'bm' not found