Cluster identification

[jupiter]

I've executed the Louvain clustering algorithm on my graph, and I got more clusters than I expected, now I would like to get more information about these clusters to understand if there are some clusters that I should not consider.

For instance, I would like to find the average local clustering coefficient for each of them using this algorithm: neo4j local clustering coefficient, the problem is that it can be executed just on the whole graph, I need to execute it for each cluster.

Actually each node is marked with a property that describes the cluster it belongs to. I tried to project part of the graph, but there is no way to project filtering on a property.

How can I do it? There is some other way to get more information about each cluster?

thanks

[mark_needham]

Hi,

I think you should be able to project part of the graph using a Cypher projection, as described here - https://neo4j.com/docs/graph-data-science/current/management-ops/cypher-projection/

That way you could restrict the projected graph based on any criteria that you like.

Cheers, Mark