Gephi “makes graphs handy”

Gephi is a tool like photoshop, but for data, the user manipulates the structures, shapes and colors to reveal hidden properties. Visual thinking with interactive interfaces is recognized to facilitate reasoning. Data analysts can use this tool to help with making hypothesis, intuitively discover patterns, isolate structure singularities or faults during data sourcing. This is a software for analyzing data sets to summarize their main characteristics with visual methods; Exploratory Data Analysis. Also use Gephi for Link Analysis, Social Network Analysis, and Biological Network Analysis.

  1. Download the tool, make sure to have Java Script dowloaded as well. Move the dowloaded Gephi.dmg file to applications. Open the application and select Open Graph File to then upload the dataset you have chosen. Here I selected one of the sample datasets provided and will model what to do after selecting your dataset. Choose Ok when the Import Report box pops up.
  2. You should now see a graph. On the left panel under the Layout module choose “Force Atlas” and click on Run to launch the algorithm. Then set the “Repulsion Strength” at 10,000 to expand the graph and press enter.

Screen Shot 2015-11-07 at 2.33.14 PM

Layouted graph

Layouted graph

3. Configure the node’s color and size through the Ranking module on the top left and choose degree and apply. To configure the color, double-click on the triangle and apply.

Screen Shot 2015-11-07 at 3.27.02 PM

4. Click on run near “Average Path Length” which is under the Statistics module on the right panel.  A setting Panel immediately appears, select Directed and click on OK. When finished the metric displays its result in a report.

Screen Shot 2015-11-07 at 3.32.03 PM

Screen Shot 2015-11-07 at 3.32.13 PM Screen Shot 2015-11-07 at 3.32.20 PM

5. Go back to the Ranking module and select “Between Centrality”. Select the diamond icon in the toolbar for size, set a min size at 10 and a max size at 50 and apply.

Screen Shot 2015-11-07 at 3.39.08 PM

6. Big nodes are overlapping smaller ones so we must go back to the Layout panel and check the box for Adjust by Sizes and run the algorithm again. The nodes will begin to disperse in order to not overlap anymore.

Screen Shot 2015-11-07 at 3.45.00 PM

7. Display the node labels by selecting the T, (when placing your mouse over T “show label” appears) set label size with the scale slider (found in the bottom, Font Size Scale appears when placing mouse over).

Screen Shot 2015-11-07 at 4.42.10 PM

8. Click on run next to Modularity line under the Statistics panel and select Randomize on the panel. Click on Ok to launch the detection, Locate the partition module on the left panel and click the refresh button.

Screen Shot 2015-11-07 at 5.22.37 PM

9. Select Modularity class, click apply. Locate the Filters module on the right panel, select the Degree Range in the Topology category and move  the slider to set its lower bound to 2. Push the filter button and now the nodes with a degree less than 2 are hidden.

Screen Shot 2015-11-07 at 5.44.24 PM

10. Preview your graph by selecting the Preview tab on the top of the banner and click on refresh. In the Node properties choose Show Labels and refresh, from preview click on SVG near Export and save your project!

Here is a video tutorial and blog tutorial that also help to guide on how to use this tool!

xgen

One Comment

  1. Genesis,

    This is a nice introduction to the layout capabilities of Gephi, and you included a number of screenshots that effectively illustrate the steps of your process. I was left a little confused about why I would want to do this, however. You chose a sample data set but it’s unclear what it was, what the various parameters you chose do and how they affect your interpretation of the results. Remember, network analysis is not appropriate for all types of data, and metrics like “betweenness centrality” might not make sense depending on what your network is graphing.

    You also want to make sure that your instructions are for Mac install and requires Java, not Java Script (remember they are, confusingly, not related). The easiest thing to do in cases like this is point to the system requirements page, and just be a little more verbose in describing why you are doing each step.

    Despite these issues, this is a solid tutorial on configuring and styling a graph in Gephi to highlight patterns in the data, and your links to further resources are very helpful.

Leave a Reply

Your email address will not be published. Required fields are marked *