I am trying to perform clustering on a graph of customers. I can use either ml based KNN or a GDS algorithm, such as Louvain community detection.How does relationships matter ?
Hi @ravitadela ,
Relationships really matters if you want to give an specific weight in a Louvain algorithm , for example if you have a graph with customer and some demographic data as Job Type and every customer are related to Job Type with a relationship and the relation have the job income, you can use this information in the Algortihm and see whats the diference usin or not relationships weight. You can try it with Bloom and the GDS capability
Even you can explore the K-Means Algorithm in Neo4j GDS , where the relationship are just used to filter de nodes.
Greetings