RESULT OBTAINED
Decision concept
Communities: detecting the groups of actors that structure the ecosystem
What must be established
A useful analysis keeps its controls visible.
- 01
Specify the type of relation and the period of the graph.
- 02
Compare computed detection and business interpretation.
- 03
Check the stability of the groups when incomplete data changes.
Decision reference
Make the next choice more explicit.
DECISION SUPPORTED
Which groups of actors are more densely connected to one another than to the rest of the ecosystem?
DELIVERABLE TO KEEP
Interpretable groups, their bridging actors and their evolution over time.
NEXT ELEMENT TO EXAMINE
DependenciesExplore →Build the graph before computing
Define the nodes, the direction and weight of the relations, the period, the sources and the deduplication rule. A relation can represent a contract, a citation, an investment or a flow, but these objects are not interchangeable. Missing links due to incomplete collection remain flagged.
- Unit of the nodes
- Type and direction of the link
- Time window
- Missing data
Distinguish community and betweenness
Community detection groups nodes according to a declared function, for example modularity. The betweenness of a node measures its presence on the shortest paths according to a specified formula and normalisation. The file displayed is a simulated illustrative set: its scores describe no real ecosystem.
- Algorithm and version
- Optimised function
- Directed or undirected graph
- Score normalisation
Test stability and business interpretation
Recompute after removing uncertain links, varying the window and changing the resolution. A group is interpreted only if it remains sufficiently stable and if external evidence explains the relations. The decision is reviewed when the coverage of the graph, the definition of the link or the membership of the bridging actors changes.
- Sensitivity analysis
- Stability of the groups
- Business validation
- Dated review
24 · ORIGINAL ASSET
Communities and bridging actors
Original graph
Three groups are distinguished by their relations. Betweenness centrality identifies the actors that connect otherwise separate communities.
Regulation
Technology
Market
Institute AConsultancy CNormalised betweenness
Teaching case · simulated data · no real-world market conclusion.
- Blondel et al., Fast unfolding of communities in large networks, 2008 (opens in a new tab)Method paper on community detection by modularity optimisation and its conventions.
- Brandes, A Faster Algorithm for Betweenness Centrality, 2001 (opens in a new tab)Algorithmic reference for defining and computing betweenness in a network.
MOVE FROM ANALYSIS TO YOUR DECISION
Produce a testable decision mandate
MOVE FROM ANALYSIS TO YOUR DECISION
Produce a testable decision mandate
RETURN TO THE DECISION SYSTEM
See the canonical definition of marketing strategy→