Function for plotting network objects using ggplot2, now replaced by the ggnet2 function, which provides additional control over plotting parameters. Please visit https://github.com/briatte/ggnet for the latest version of ggnet2, and https://briatte.github.io/ggnet/ for a vignette that contains many examples and explanations.

ggnet(
  net,
  mode = "fruchtermanreingold",
  layout.par = NULL,
  layout.exp = 0,
  size = 9,
  alpha = 1,
  weight = "none",
  weight.legend = NA,
  weight.method = weight,
  weight.min = NA,
  weight.max = NA,
  weight.cut = FALSE,
  group = NULL,
  group.legend = NA,
  node.group = group,
  node.color = NULL,
  node.alpha = alpha,
  segment.alpha = alpha,
  segment.color = "grey50",
  segment.label = NULL,
  segment.size = 0.25,
  arrow.size = 0,
  arrow.gap = 0,
  arrow.type = "closed",
  label = FALSE,
  label.nodes = label,
  label.size = size/2,
  label.trim = FALSE,
  legend.size = 9,
  legend.position = "right",
  names = c("", ""),
  quantize.weights = FALSE,
  subset.threshold = 0,
  top8.nodes = FALSE,
  trim.labels = FALSE,
  ...
)

Arguments

net

an object of class network, or any object that can be coerced to this class, such as an adjacency or incidence matrix, or an edge list: see edgeset.constructors and network for details. If the object is of class igraph and the intergraph package is installed, it will be used to convert the object: see asNetwork for details.

mode

a placement method from those provided in the sna package: see gplot.layout for details. Also accepts the names of two numeric vertex attributes of net, or a matrix of numeric coordinates, in which case the first two columns of the matrix are used. Defaults to the Fruchterman-Reingold force-directed algorithm.

layout.par

options to be passed to the placement method, as listed in gplot.layout. Defaults to NULL.

layout.exp

a multiplier to expand the horizontal axis if node labels get clipped: see expand_range for details. Defaults to 0 (no expansion).

size

size of the network nodes. If the nodes are weighted, their area is proportionally scaled up to the size set by size. Defaults to 9.

alpha

a level of transparency for nodes, vertices and arrows. Defaults to 1.

weight

the weighting method for the nodes, which might be a vertex attribute or a vector of size values. Also accepts "indegree", "outdegree", "degree" or "freeman" to size the nodes by their unweighted degree centrality ("degree" and "freeman" are equivalent): see degree for details. All node weights must be positive. Defaults to "none" (no weighting).

weight.legend

the name to assign to the legend created by weight. Defaults to NA (no name).

weight.method

see weight

weight.min

whether to subset the network to nodes with a minimum size, based on the values of weight. Defaults to NA (preserves all nodes).

weight.max

whether to subset the network to nodes with a maximum size, based on the values of weight. Defaults to NA (preserves all nodes).

weight.cut

whether to cut the size of the nodes into a certain number of quantiles. Accepts TRUE, which tries to cut the sizes into quartiles, or any positive numeric value, which tries to cut the sizes into that many quantiles. If the size of the nodes do not contain the specified number of distinct quantiles, the largest possible number is used. See quantile and cut for details. Defaults to FALSE (does nothing).

group

the groups of the nodes, either as a vector of values or as a vertex attribute. If set to mode on a bipartite network, the nodes will be grouped as "actor" if they belong to the primary mode and "event" if they belong to the secondary mode.

group.legend

the name to assign to the legend created by group.

node.group

see group

node.color

a vector of character strings to color the nodes with, holding as many colors as there are levels in node.group. Defaults to NULL, which will assign grayscale colors to each group.

node.alpha

transparency of the nodes. Inherits from alpha.

segment.alpha

the level of transparency of the edges. Defaults to alpha, which defaults to 1.

segment.color

the color of the edges, as a color value, a vector of color values, or as an edge attribute containing color values. Defaults to "grey50".

segment.label

the labels to plot at the middle of the edges, as a single value, a vector of values, or as an edge attribute. Defaults to NULL (no edge labels).

segment.size

the size of the edges, in points, as a single numeric value, a vector of values, or as an edge attribute. Defaults to 0.25.

arrow.size

the size of the arrows for directed network edges, in points. See arrow for details. Defaults to 0 (no arrows).

arrow.gap

a setting aimed at improving the display of edge arrows by plotting slightly shorter edges. Accepts any value between 0 and 1, where a value of 0.05 will generally achieve good results when the size of the nodes is reasonably small. Defaults to 0 (no shortening).

arrow.type

the type of the arrows for directed network edges. See arrow for details. Defaults to "closed".

label

whether to label the nodes. If set to TRUE, nodes are labeled with their vertex names. If set to a vector that contains as many elements as there are nodes in net, nodes are labeled with these. If set to any other vector of values, the nodes are labeled only when their vertex name matches one of these values. Defaults to FALSE (no labels).

label.nodes

see label

label.size

the size of the node labels, in points, as a numeric value, a vector of numeric values, or as a vertex attribute containing numeric values. Defaults to size / 2 (half the maximum node size), which defaults to 6.

label.trim

whether to apply some trimming to the node labels. Accepts any function that can process a character vector, or a strictly positive numeric value, in which case the labels are trimmed to a fixed-length substring of that length: see substr for details. Defaults to FALSE (does nothing).

legend.size

the size of the legend symbols and text, in points. Defaults to 9.

legend.position

the location of the plot legend(s). Accepts all legend.position values supported by theme. Defaults to "right".

names

deprecated: see group.legend and size.legend

quantize.weights

deprecated: see weight.cut

subset.threshold

deprecated: see weight.min

top8.nodes

deprecated: this functionality was experimental and has been removed entirely from ggnet

trim.labels

deprecated: see label.trim

...

other arguments passed to the geom_text object that sets the node labels: see geom_text for details.

Details

The degree centrality measures that can be produced through the weight argument will take the directedness of the network into account, but will be unweighted. To compute weighted network measures, see the tnet package by Tore Opsahl (help("tnet", package = "tnet")).

See also

ggnet2 in this package, gplot in the sna package, and plot.network in the network package

Author

Moritz Marbach and Francois Briatte, with help from Heike Hofmann, Pedro Jordano and Ming-Yu Liu

Examples

# Small function to display plots only if it's interactive
p_ <- GGally::print_if_interactive

library(network)
#> 
#> ‘network’ 1.18.2 (2023-12-04), part of the Statnet Project
#> * ‘news(package="network")’ for changes since last version
#> * ‘citation("network")’ for citation information
#> * ‘https://statnet.org’ for help, support, and other information
#> 
#> Attaching package: ‘network’
#> The following object is masked from ‘package:Hmisc’:
#> 
#>     is.discrete

# random adjacency matrix
x           <- 10
ndyads      <- x * (x - 1)
density     <- x / ndyads
m           <- matrix(0, nrow = x, ncol = x)
dimnames(m) <- list(letters[1:x], letters[1:x])
m[row(m) != col(m)] <- runif(ndyads) < density
m
#>   a b c d e f g h i j
#> a 0 0 0 0 0 0 0 1 0 1
#> b 0 0 0 0 0 0 0 0 0 0
#> c 0 0 0 0 0 0 0 0 0 1
#> d 0 0 0 0 0 0 0 0 0 1
#> e 0 0 0 0 0 0 0 0 0 0
#> f 0 0 0 0 0 0 0 1 0 0
#> g 0 0 0 1 0 0 0 1 0 0
#> h 0 0 0 0 0 0 0 0 0 0
#> i 0 0 0 0 1 0 0 0 0 0
#> j 1 0 0 1 0 0 0 0 0 0

# random undirected network
n <- network::network(m, directed = FALSE)
n
#>  Network attributes:
#>   vertices = 10 
#>   directed = FALSE 
#>   hyper = FALSE 
#>   loops = FALSE 
#>   multiple = FALSE 
#>   bipartite = FALSE 
#>   total edges= 8 
#>     missing edges= 0 
#>     non-missing edges= 8 
#> 
#>  Vertex attribute names: 
#>     vertex.names 
#> 
#> No edge attributes

ggnet(n, label = TRUE, alpha = 1, color = "white", segment.color = "black")


# random groups
g <- sample(letters[1:3], 10, replace = TRUE)
g
#>  [1] "b" "a" "a" "a" "b" "b" "a" "b" "a" "b"

# color palette
p <- c("a" = "steelblue", "b" = "forestgreen", "c" = "tomato")

p_(ggnet(n, node.group = g, node.color = p, label = TRUE, color = "white"))
#> Warning: node groups and colors are of unequal length; using grayscale colors


# edge arrows on a directed network
p_(ggnet(network(m, directed = TRUE), arrow.gap = 0.05, arrow.size = 10))
#> Warning: Removed 1 rows containing missing values (`geom_segment()`).