Exploring Added Network Connectivity with subgrouping Strategies by NPollesch
ORDNetwork explores three different subgrouping strategies for how to optimally group members of a heirarchical organization so that the new groups minimize total distance in the organization.
Subgroups were created uniformly randomly, using a greedy algorithm (maximum person to person distance), and using an anti-greedy algorithm (minimum person to person distance) as a comparison
Initial results showed that random grouping was nearly as effective as the greedy algorithm and was significantly easier to implement. Both methods performed better than the minimum distance algorithm,
For the actual grouping exercise that motivated the project, a random algorithm was paired with two metrics for evenness to ensure group distance was minimized and a mixture of upper and lower management was achieved in each subgroup.
See the "Mix the managers" version of the .docx for a project summary.