Chapter 3 ggraph

We’ll make use of the packages ggraph and tidygraph which build on the ggplot2 framework. This is useful, because ggplot2 has a large community, and almost all imaginable questions have already been answered on websites like stackoverflow.

Great documentation exists for:

Here, here, and here are some further useful introduction to visualizing networks.

3.1 Why not {insert your favourite network visualization software/package}?

A sensible question is, why not use something like gephi or UCInet? While these programs are (much?) easier to use than R, they are point-and-click. This has advantages: it’s easy to use and designed for networks only. They have disadvantages: they are typically focussed on visualizing one network (what if you have to visualize hundreds?), and they do not favour reproducibility; if you forgot a node, you’d have to do your point and click thingy all over again. Moreover, often you want to make more standard visualizations (e.g., histogram of betweenness of nodes) which you can do in R, because you can do almost anything in it.

Why not R-packages like sna or igraph? I am highly biased: I know ggplot2 well, and ggraph builds on this system. There is huge support for ggplot2 and thus indirectly for ggraph. The strength of sna and igraph is probably more through their analyses capabilities (in fact, ggraph borrows these capabilities from these packages).