Cluster Runaways

magic_al

Hi guys,

I'm looking for a method to successfully detect cluster runaways. With that I mean objects which flee from cluster A into cluster B considering different periodes of observation. Is that posible with RapidMinder? I aim for spatial trend analysis over time.

MariusHelf

There are several possibilities. In any case you'll need an id attribute in your data such that you can uniquely identify an object. Then you could e.g. cluster your data at different points of time, join it together via the id attribute and compare the assigned clusters. If they differ, you have a runaway.

Best regards,
Marius