CODEWITHSUNDEEP
rigraphggnet2
Max Stüwe
Max Stüwe

Reputation: 1

Defining Node shape in a Network plot with an additional attribute table in R

I am working on plotting a Network and it contains two different types of Nodes which I want to visualise with different shapes. For that I made an additional table in which I specified which structure is which type using a binary system. Now I want to specify in my plot function that the structures with 1 are to be triangles and the ones with 0 as circles. My data for the Network is in the format of an adjacency matrix (I use igraph) and I am using ggnet2 for the plotting of it.

this is how I imported the data:

am <- as.matrix(read.csv2("mydata.csv", header = T, row.names = 1))
g <- graph_from_adjacency_matrix(am, mode = "undirected")
attr <- read.csv2("myattributes.csv", header = T, row.names = 1)

this is how I would plot it but I dont know how to specify the shape function

ggnet2(g, size = "degree", node.color = "darkgreen", shape = ??????)

Thanks in advance for your help!

Upvotes: 0

Views: 229

Answers (1)

nJGL
nJGL

Reputation: 864

Note that the package-requirements for plotting igraphs with ggnet2 include ggplot2, sna and network as well as intergraph as a bridge.

ggnet2 is prettier, sure, but the igraph-way is this:

g <- erdos.renyi.game(100,100,'gnm')
V(g)$shape <- sample(c('csquare','circle'), 100, replace=T)
plot(g, vertex.label = NA)

Note that I added two igraph-style shapes as vertex-attributes to g above. In ggent2 you can provide a vector with shapes, but they can be any values (even a factor), or numbers (the usual gray circle is 19. Try this out to plot in ggnet2

ggnet2(g, shape=19)
ggnet2(g, shape=10+round(1:100/10))
ggnet2(g, shape=factor(V(g)$shape))
V(g)$shape <- sample(c('One shape','Another shape'), 100, replace=T)
ggnet2(g, shape=V(g)$shape, size = "degree", node.color = "darkgreen")

Note that, if you add attributes to your vertices after separately loading attribute data (as you do above), it may be so that the very order of your data matters. Make sure your table import actually works as intended with the correct attribute being assigned to the correct vertex. I find it a good practice to tie all values as attributes on the igraph-object (edge- and vertex attributes alike) rather than letting the network data live in different dataframes or loose vectors to be combined in order to correctly visualise a network.

Upvotes: 0

Related Questions