If you are, this is some interesting reading for you:
I am looking for references on reverse clustering. It consists in identifying actionable metrics that allow you to separate groups detected via non-actionable metrics.
Example:
On a remote planet, we have 6 types of aliens: A, B, C, D, E, F. It is very easy to identify these groups (that is, assign a cluster to each individual) using non-actionable metrics: each group has its own religion (F does not have any). So religion is the standard, non-actionable metric used for clustering.
Now A and B are at war. B wants to identify metrics to disriminate between A and B, that are
- Actionable: can be leveraged for tactical warfare
- Discriminate very well between A, B, and C-F
- Will not cause problems with groups C-F
Here, A does not have friends, so attacking A will not cause C-F to be upset. But in some other contexts, one must be very careful about reactions from groups C-F. An actionable metric that was identified is the following: individuals from group A tend ...
Read full article at https://bit.ly/10zuA1r
You may leave the list at any time by sending the command
SIGNOFF allstat
to [log in to unmask], leaving the subject line blank.
|