centdf: Get desired centralities in a dataframe

In kturetsky/sneasy: Helper Functions for Social Network Analysis

Get desired centralities in a dataframe

Description

Calculate common node-level centrality measures on a network.

Usage

centdf(g, pps, types = "all", norm = "none", inout = TRUE, include_atts = TRUE)

Arguments

g	igraph graph
pps	An optional vector of participant ids to include in the dataframe so that non-participants' centrality scores are not included. If argument is missing, a warning message will be issued that non-participants may be included. To suppress warning, use pps=NULL.
types	Character string; which centralities to calculate (default "all"; other valid values are: "alldisconnect" (all suitable for disconnected graphs, i.e., not closeness or eigenvector centrality), "degree"/"deg", "betweenness"/"btwn", "closeness"/"close", "harmonic"/"hclose", "strength"/"str", "avgstrength"/"avgstr", "eigenvector"/"eigen").
norm	Character string; which centralities to normalize if applicable (default "none"; otherwise, takes the same values as 'types'). Normalization is available for degree, betweenness, closeness, and harmonic closeness.
inout	Logical; if TRUE (default), will return in, out, and total centrality of each type if applicable. If FALSE, will only return total centrality. Only total centrality will be calculated if the graph object g is undirected. Additionally, betweeness centrality does not have an in- or out-type so the function will always calculate one value for betweenness.
include_atts	Logical; if TRUE (default), will include node attributes in returned dataframe

Details

Centrality measures will be calculated according to the type of network provided; if the graph is directed and weighted, centrality measures will be calculated using directed ties and weights, as applicable (except for eigenvector centrality, which is not well-defined for directed graphs and will thus convert the network to an undirected graph prior to calculation). If undirected and/or unweighted centrality calculations are desired, send the function an undirected and/or unweighted graph.

Closeness and eigenvector centralities are not well-defined for disconnected graphs. Asking for these centrality measures to be calculated on a graph with disconnected components will produce a warning. Harmonic centrality is recommended in place of closeness centrality for disconnected graphs.

If a particular node has no ties, their average strength (total strength of ties divided by number of ties) will be returned as 0 to indicate no strength (not NaN as dividing by 0 ties would produce).

Value

A dataframe of centralities (and, optional, node attributes) for each individual

See Also

igraph::degree(), igraph::betweenness(), igraph::closeness, CINNA::harmonic_centrality(), igraph::strength(), igraph::eigen_centrality()

kturetsky/sneasy documentation built on Aug. 29, 2022, 7:11 a.m.