This demo illustrates the power of selective filtering to provide focus on a subset of data. It uses the graph's kCores to show which nodes have a higher degree of 'connectedness' than others.

Filtering

The opening state of the graph shows the full dataset; all suspects in all countries are displayed. The result is a typically complex and hard to understand network.

By controlling the displayed data, the user can make decisions on information they are interested in by filtering out certain countries or nodes with low 'connectedness'.

kCores

The kCores parameter can be adjusted to show only those nodes (individuals) that are connected to at least a defined number of other nodes. For example, a setting of '1' will filter all nodes that have less than 1 connection to another node, so all the 'singletons' will be hidden. A setting of '4' will hide all nodes that only have connections to 3 or fewer nodes, and will repeat this until all remaining nodes have at least 4 connections. Try setting kCores to '8' to see the dramatic simplification of the graph.

It is clear that kCores is a powerful way to show only more connected nodes; by sequentially increasing the parameter the user can explore the relations of highly connected nodes which are typically significant in a network.

Key functions used:

• chart.filter
• chart.graph().kCores