I have 114 vectors with 6 boolean attributes. I saw that might be several distinct clusters in a simple visualization. K-means clustering on the transformed vectors (true = 1, false = 0) results in roughly the clusters that I had seen in the visualization.

However, I am not sure what the most appropriate clustering method for this kind of data is, and how to determine the confidence in those factors (the k-means results change every time due to randomization). Should I treat the data as nominal or as numerical data?

What would be the best way to do a cluster analysis on this kind of data in R?

  • $\begingroup$ You could also try one of the hierarchical methods with an appropriate measure of similarity/dissimilarity. For tha latter, Sokal-Michener, Sokal-Sneath, Rogers-Tinimoto, Russell-Raova, Jaccard, Czekanowski, Sokal-Sneath #2 measures of similarity comes to mind. $\endgroup$ – Roman Luštrik Apr 29 '12 at 6:36
  • $\begingroup$ stats.stackexchange.com/questions/6252/… $\endgroup$ – Etienne Low-Décarie Apr 30 '12 at 1:04

I would have a glance to the mona function in the cluster package. It seems to address your question.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.