Fig. 2

The methodology used with data from the Seattle Police Department crime event data (taken from data.gov) where the type of ‘Crime’ and the ‘Precinct’ are used. a Shows the cluster entropy trace for the window size set. b Shows the walks of length 2 between labels (triangulation) of the variable which is effectively the sharing of label values in events. For this example this corresponds to crime types having been reported in the same precinct (left/blue edges) and for different precincts to have an edge they have at least 1 common crime reported. c Has 3 different edge types for the walks of length 2 between variable labels and the length 1 of the number of label co-occurrence counts between different variables