Network Visualization in the 'ggplot2' Framework

Network visualization in the 'ggplot2' framework. Network functionality is provided in a single 'ggplot2' layer by calling the geom 'net'. Layouts are calculated using the 'sna' package, example networks are included.

Sam Tyner, Heike Hofmann 2016-12-08

geomnet is a package built on top of the most recent major ggplot2 release. It provides a ggplot2 geom called geom_net to visualize graphs and networks. It also include the function stat_net to calculate network layouts with the sna package. Finally, the function geom_circle is included to draw circles using ggplot2.

You can install geomnet directly from CRAN install.packages('geomnet') or from Github devtools::install_github("sctyner/geomnet")

Examples

ggplot2 Theme Elements

This example shows the theme inheritance properties of the theme elements of ggplot2. Note: this example has not been updated since the release of ggplot2 2.2.0 and as such the content may have changed.

library(dplyr)
library(geomnet)
data(theme_elements)
TEnet <- fortify(as.edgedf(theme_elements$edges[,c("parent", "child")]), theme_elements$vertices)
# create a degree variable for use later
TEnet <- TEnet %>%
  group_by(from_id) %>%
  mutate(degree = sqrt(10 * n() + 1))
# plot
ggplot(data = TEnet,
       aes(from_id = from_id, to_id = to_id)) +
  geom_net(layout.alg = "fruchtermanreingold",
    aes(fontsize = degree), directed = TRUE,
    labelon = TRUE, size = 1, labelcolour = 'black',
    ecolour = "grey70", arrowsize = 0.5,
    linewidth = 0.5, repel = TRUE) +
  theme_net() +
  xlim(c(-0.05, 1.05))

Blood Donation Diagram

In this example, we reimagine the traditional blood donation diagram as a directed network. Arrows point to the blood type that receives. This example provides two data frames to geom_net().

library(geomnet)
data(blood)
ggplot(data = blood$edges, aes(from_id = from, to_id = to)) +
  geom_net(colour = "darkred", layout.alg = "circle", labelon = TRUE, 
           size = 15, directed = TRUE, vjust = 0.5, labelcolour = "grey80",
           arrowsize = 1.5, linewidth = 0.5, arrowgap = 0.05,
           selfloops = TRUE, ecolour = "grey40") + 
  theme_net() 

Harry Potter Peer Support Network

In this fun example from this website, there is a tie between two students if one provides emotional support to the other at some point in the book. It is a directed network, so in the visualization, the arrow points to the student receiving support.

library(geomnet)
head(hp.edges)
#>          name1              name2 book
#> 1  Dean Thomas Harry James Potter    1
#> 2  Dean Thomas   Hermione Granger    1
#> 3  Dean Thomas Neville Longbottom    1
#> 4  Dean Thomas     Ronald Weasley    1
#> 5  Dean Thomas    Seamus Finnigan    1
#> 6 Fred Weasley     George Weasley    1
head(hp.chars)
#>                name schoolyear gender      house
#> 1      Adrian Pucey       1989      M  Slytherin
#> 2    Alicia Spinnet       1989      F Gryffindor
#> 3  Angelina Johnson       1989      F Gryffindor
#> 4 Anthony Goldstein       1991      M  Ravenclaw
#> 5     Blaise Zabini       1991      M  Slytherin
#> 6     C. Warrington       1989      M  Slytherin
hp.all <- fortify(as.edgedf(hp.edges), hp.chars, group = "book")
#> Using name1 as the from node column and name2 as the to node column.
#> If this is not correct, rewrite dat so that the first 2 columns are from and to node, respectively.
#> Joining edge and node information by from_id and name respectively.
# only plot the characters with any connections in a given book. 
ggplot(data=hp.all, aes(from_id = from, to_id = to_id)) + 
  geom_net(fiteach=T, directed = T, size = 3, linewidth = .5, 
           ealpha = .5, labelon = T, fontsize = 3, repel = T, 
           labelcolour = "black", arrowsize = .5, singletons = FALSE,
           aes(colour = house, group = house, shape = gender)) + 
  scale_colour_manual(values = c("#941B08","#F1F31C", 
                                 "#071A80", "#154C07")) + 
  facet_wrap(~book, labeller = "label_both") + 
  theme_net() + theme(panel.background = element_rect(colour = 'black'))

Interactivity with plotly!

Now including interactivity with ggplotly()!

library(geomnet)
# devtools::install_github("ropensci/plotly")
library(plotly)
data("football")
# data step: merge vertices and edges
ftnet <- fortify(as.edgedf(football$edges), football$vertices)
 
# create data plot
gg <- ggplot(data = ftnet,
       aes(from_id = from_id, to_id = to_id)) +
  geom_net(layout.alg = 'fruchtermanreingold',
    aes(colour = value, group = value,
        linetype = factor(same.conf != 1)),
    linewidth = 0.5,
    size = 5, vjust = -0.75, alpha = 0.3) +
  theme_net() +
  theme(legend.position = "bottom") +
  scale_colour_brewer("Conference", palette = "Paired")  +
  guides(linetype = FALSE)
ggplotly(gg)