Section 3 – Network Visualization and Analysis: Case Study

Lesson 8 – Creating High-Resolution Print-Ready Visualizations

In the previous lessons we created graph visualizations and demonstrated various techniques that can be used to better understand the data. Those visualizations look good enough and informative for analysis, however, if you want to make them presentation-ready or prepare them for high-resolution print, we need to make a few extra steps to save those graphs in higher quality formats. In this lesson we will demonstrate how you can do that.

At the moment our graph looks something like this:

There are several steps we can take to make this visualization look less messy:

1) First, let’s emphasize bigger nodes, so they are more noticeable in relation to the smaller ones. In order to do that, open the “Ranking” pane and click “Spline”. There you can alter the shape of the curve that maps sizes of nodes to the metrics you’re using (e.g. betweenness centrality or degree). Drag the top part of the curve to the left to make  node sizes increase faster.


2) Let’s then push the nodes apart from each other more, so they don’t intersect too much. For that, you need to open “Layout” pane >> Force Atlas >> and increase “Repulsion strength”, decrease the “Gravity” parameter and click Apply again (make sure that “Adjust by Sizes” option is also selected, so the Force Atlas algorithm takes the note of the nodes’ sizes.


3) We still have some node labels overlapping. In order to avoid that, choose “Label Adjust” option in Layout and click apply again.


4) Finally, now that we have an image that is readable and clear, we can convert it into print-ready format. In order to do that, choose the “Preview” tab at the top left menu of Gephi. Then in Preview Settings select: Show Labels, Label Font: 24 (can be higher or lower than that), select “Curved” in the Undirected list (so that the edges are curved) and then click “Refresh”:

Screen Shot 2016-06-30 at 14.47.57

5) Some adjustments can still be made here. For instance, we could make the edges thicker (set “Thickness” to “5” instead of “1” – In “Undirected” list as our graph is undirected). We also could set the size of the node label font a bit smaller (from 24 to 18, so it’s not so messy) and get something like that:


6) Finally, we can also change the background of the graph to black (at the bottom of the window)

Screen Shot 2016-06-30 at 14.56.28

And also – switch the label border to “On” and choose Label Border Color as “Parent” so it’s the same color as the graph. This way we get a graph that looks more like a Tag Cloud, but a much more advanced and context-aware version of it. All the nodes are aligned depending on how densely connected they are:


You could, of course, switch the labels off completely, then you get a more abstract representation of the whole graph: