gapply {sna}

Apply Functions Over Vertex Neighborhoods

Description

Returns a vector or array or list of values obtained by applying a function to vertex neighborhoods of a given order.

Usage

gapply(X, MARGIN, STATS, FUN, ..., mode = "digraph", diag = FALSE, 
    distance = 1, thresh = 0, simplify = TRUE)

Arguments

X

one or more input graphs.

MARGIN

a vector giving the “margin” of X to be used in calculating neighborhoods. 1 indicates rows (out-neighbors), 2 indicates columns (in-neighbors), and c(1,2) indicates rows and columns (total neighborhood).

STATS

the vector or matrix of vertex statistics to be used.

FUN

the function to be applied. In the case of operators, the function name must be quoted.

...

additional arguments to FUN.

mode

"graph" if X is a simple graph, else "digraph".

diag

boolean; are the diagonals of X meaningful?

distance

the maximum geodesic distance at which neighborhoods are to be taken. 1 signifies first-order neighborhoods, 2 signifies second-order neighborhoods, etc.

thresh

the threshold to be used in dichotomizing X.

simplify

boolean; should we attempt to coerce output to a vector if possible?

Details

For each vertex in X, gapply first identifies all members of the relevant neighborhood (as determined by MARGIN and distance) and pulls the rows of STATS associated with each. FUN is then applied to this collection of values. This provides a very quick and easy way to answer questions like:

• How many persons are in each ego's 3rd-order neighborhood?

• What fraction of each ego's alters are female?

• What is the mean income for each ego's trading partners?

• etc.

With clever use of FUN and STATS, a wide range of functionality can be obtained.

Value

The result of the iterated application of FUN to each vertex neighborhood's STATS.

Author(s)

Carter T. Butts [e-mail]

See Also

apply, sapply

Examples

#Generate a random graph
g<-rgraph(6)

#Calculate the degree of g using gapply
all(gapply(g,1,rep(1,6),sum)==degree(g,cmode="outdegree"))
all(gapply(g,2,rep(1,6),sum)==degree(g,cmode="indegree"))
all(gapply(g,c(1,2),rep(1,6),sum)==degree(symmetrize(g),cmode="freeman")/2)

#Find first and second order neighborhood means on some variable
gapply(g,c(1,2),1:6,mean)
gapply(g,c(1,2),1:6,mean,distance=2)