"How to calculate Silhouette validation for k-means using R in Rapidminer?"

xixirhwfy

Hello everyone,
  I'm trying to use Silhouette validation for k-means clustering, but there's no such operator. So I'm trying to add R "execute script" operator to calculate silhouette. But I don't know much about R language. I want to calculate k-means clustering using operator in Rapidminer and only do the silhouette analysis in the R script. Does anyone know how to write such script? And how to connect the Rapidminer operator together with the R operator to make the whole process work and how the parameter keys are set? Better to make an example using XML, thank you!

awchisholm

Hello,

How about this?

http://rapidminernotes.blogspot.co.uk/2011/06/counting-clusters-part-r.html

Andrew

xixirhwfy

Thank you, Andrew! But I want to use kmeans clustering not the k-medoid(pam()?), and I want to use the operator in Rapidminer to calculate kmeans, then connect to a "R:execute script" operator which only calculate the Silhouette value, could you please make another simpler example to show this? Because I know nothing about the R language. Thank you very much!

awchisholm

Hello,

As it happens, I'm contributing a chapter to an upcoming book - http://rapidminerbook.com - that does something even closer to what you want. I'm afraid you'll have to wait for that. You could have a go yourself of course :-)

Andrew