igraph and SNA: an amateur’s dabbling

I’ve been playing with the igraph package a bit lately (see previous post HERE) and wanted to approach a problem I once visited in the past. The basic gist of the problem is this:

Students in a class are asked their top three favorite students to work with (rank order).  After a social intervention this same question is posed again to students.  The intended outcome of the intervention is that the distribution of students receiving many or very few choices will diminish.  In other words the dorks will become less dorky and the popular students will become less popular.  The idea is to visual this relationship.

Here is a script of one such visualization.  It’s a bit light on annotations but merely experimenting with the code should give a good sense of what is occurring.

#create a data set
X <-lapply(1:10, function(i) sample(LETTERS[c(1:10)[-i]], 3))
Y <- data.frame(person = LETTERS[1:10], sex = rbinom(10, 1, .5), do.call(rbind, X))
names(Y)[3:5] <- paste0("choice.", 1:3)

#reshape the data to long format
Z <- reshape(Y, direction="long", varying=3:5)
colnames(Z)[3:4] <- c("choice.no",  "choice")
rownames(Z) <- NULL
Z <- Z[, c(1, 4, 3, 2)]

#turn the data into a graph structure
edges <- as.matrix(Z[, 1:2])
g <- graph.data.frame(edges, directed=TRUE)
V(g)$label <- V(g)$name

#change label size based on number of votes
SUMS <- data.frame(table(Z$choice))
SUMS$Var1 <- as.character(SUMS$Var1)
SUMS <- SUMS[order(as.character(SUMS$Var1)), ]
SUMS$Freq <- as.integer(SUMS$Freq)
label.size <- 2
V(g)$label.cex <- log(scale(SUMS$Freq) + max(abs(scale(SUMS$Freq)))+ label.size)

#Color edges that are reciprocal red
x <- t(apply(edges, 1, sort))
x <- paste0(x[, 1], x[, 2])
y <- x[duplicated(x)]
COLS <- ifelse(x %in% y, "red", "gray40")
E(g)$color <- COLS

#reverse score the choices.no and weight
E(g)$width <- (4 - Z$choice.no)*2

#color vertex based on sex
V(g)$gender <- Y$sex
V(g)$color <- ifelse(V(g)$gender==0, "pink", "lightblue")

#plot it
opar <- par()$mar; par(mar=rep(0, 4)) #Give the graph lots of room
plot.igraph(g, layout=layout.auto(g))

For an additional script of this analysis with 20 students click here.

For helpful igraph documentation click here


About tylerrinker

Data Scientist and Open-source developer/contributor with a bent for the quantitative and a passion for text analysis.
This entry was posted in igraph and tagged , , , , , . Bookmark the permalink.

4 Responses to igraph and SNA: an amateur’s dabbling

  1. Juanluco says:

    Firstly I would like to congratulate for your post! But in your code have a dumb mistake: in the code of the post appears twice paste0 and I guess that it is paste. I think that if you want to visualize a graph you only need plot(g). Anyway this post is very interesting. Thanks!

  2. tylerrinker says:

    Thank you for the feedback. As of R 2.15 paste0 is a function that is shorthand for paste(…., sep=””). And yes plot will work instead of plot.igraph.

  3. tylerrinker says:

    Looking at the graph it would also be nice to vary arrow head size with the width of the edge.

  4. Pingback: …a scientific crowd | FreshBiostats

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s