I have a dataset with 6 entities, each entity has 20 attributes. I am investigating the cluster structure considering that the 6 entities shares some attributes. The clustering method that I am using is the Louvain resolution method.
There are 3 attributes that are shared by all the entities. I expected these attributes to influence the cluster structure. But in the final cluster, these shared attributes are shown as isolated nodes. So characterized by a degree=0. Could you please explain how the degree is calculated in this case?
Not sure to understand well what you are describing. Basically, you should have nodes (keywords, entities…) and edges (links between entities often based on a proximity measure).
Degree and consequently isolated nodes are directly affected by:
- The proximity measure used to build the edges
- The filters on edges used
I hope it helps!