I need to analyse questionnaire survey data with mixed data types (nominal, ordinal, continuous). I want to cluster the variables. So far I only have dead ends.

  • I know I can use daisy in the cluster package to cluster the cases, but I want to cluster the variables.
  • lots of cluster procedures like ICLUST can cluster variables but they don't allow nominal data types.

The only thing I can think of is to split up the categorial variables into binary dummy variables, standardise everything, and then apply ICLUST.

  • 1
    $\begingroup$ Use Gower measure. About measures nominal data, read this. For hierarchical and some other methods of clustering, it makes no difference whether you cluster individuals or attributes. $\endgroup$ – ttnphns Apr 3 '14 at 12:27

The R package polLCA can handle all different datatypes almost seamlessly via latent cluster analysis.

  • 3
    $\begingroup$ Can you say more about this package, & latent cluster analysis? How would they help resolve the OP's problem? $\endgroup$ – gung - Reinstate Monica Aug 6 '15 at 5:17

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