DBCSN cluster size

I have been running a DBSCAN algorithm on R for a project where I'm creating clusters of individuals based on the location they are in.

I need clusters of 3 people (k=3), each matched on location each individual and my eps value is 0.5.

The problem I'm facing with my algorithm is that the cohorts/clusters are of variable size, and can be huge.

Is there any viable way for me to split up each of these large clusters (ie, for example, if a cohort/cluster is more than 6 people I want to split them into smaller cohorts/clusters such that each one contains between 3 and 6 people)

Thanks!

https://stackoverflow.com/questions/67040587/dbcsn-cluster-size April 11, 2021 at 09:04AM