By placing Degree Range in the Queries and adjusting the Degree range to begin at “2” I eliminated leaves (nodes with only one degree in the data). This is done in the Appearance window.įiltering is available by dragging options from the right side Filters window into the Queries box.

Modularity, to identify “communities” of nodes located close to one another, based on a chosen data metric (Once the “Modularity Class” is constructed the user can apply color by community to visualize the groupings identified). I ran the following statistics on my data:Įdge Overview-Average Path Length, to identify how close nodes are to one anotherĭiameter, to understand the distance between the two furthest nodes in the graphīetweeness Centrality, to identify which nodes are most centrally connected in the graph Icons along the bottom of the Overview control labels display, font size of labels (including sizing according to node size), and background color of network graph (black or white). Also clicking “Adjust by Size” in the Layout window for Force Atlas keeps the nodes from overlapping once they are sized. The nodes’ size can be controlled in the ranking using Betweeness Centrality. The nodes and the edges can be colored according to degree once the metrics are run to add the degree to your data. (The user can control the scaling in the Layout window).Ĭolors can be chosen in the Appearance window using preset palettes or a color wheel to generate your own color scheme.

The Fruchterman-Reingold spread out the nodes almost equidistantly as the nodes are attracted to the center, whereas Force Atlas grouped the nodes according to attraction and repulsion scales which make the nodes cluster closer together when connections are many, and repel from each other when less connected or unconnected, making those nodes further apart. With my small datasets, these created two-dimensional layouts that easily showed all nodes and relationships. I practiced with Layout algorithms Force Atlas, Force Atlas 2, and Fruchterman-Reingold. The title column must be copied to the “Label” column in the data table view in Gephi. The CSV is imported through File, Import spreadsheet… The nodes were collected in a “title” column as one CSV file, then the edges and direction data was collected in a second data file, renaming the columns to: source (node), target (node), and type (direction).

The XML file was brought into Open Refine where it was prepared for export as a CSV. GEXF and GML are directly supported by Gephi and can be imported through File, Open… Gephi for analyzing and visualizing these graph data The datasets I selected were in GEXF, GML, and XML formats. As it turned out, my novice skillset cannot even begin to mimic these samples, but I was more attracted to this style of network graphic, rather than the massive datasets that gave more of a paint-splatter/artwork effect. All of the above examples are beautifully visualized with simple color palettes and connections that we can interpret relatively easily without needing interactive tools or multiple screen views. I was interested in datasets for this lab where the viewer can see the points and literally connect the dots.

Figure 3 The Clubs that Connect the World Cup (Aisch, 2014)

Steve ross ibm news explorer