In the early 70s Lorr had a method of clustering based on the values of
correlations or distances. r correlations for variables, and Q
correlations for entities. One specified a minimum coefficient for
inclusion in a cluster and a maximum coefficient for inclusion in a
cluster. One feature of this method was that some items would be left
unclustered or outliers. For example, Los Angeles county stayed as its
own cluster.
By any chance has anybody put this into Python?
Mcquitty had a series of methods that identified "siblings", then
"cousins", and "second cousins" and so forth. By any chance has anybody
put this into Python?