I wanted to ask anyone was aware of a type of two-stage analysis where clusters are used as a dependent variable in prediction models?

For example, suppose I had used an unsupervised model based on five categorical covariates, and I generated 3 clusters as a consequence.

Is it possible to use a representation of one of these clusters as a dependent variable in another model, to evaluate how well another set of mixed covariates would predict the cluster?

Sounds potentially outrageous, but would welcome comments and feedback.

  • 1
    $\begingroup$ Can't think of any example but not outrageous at all in my opinion. $\endgroup$
    – Erwan
    Jul 29, 2021 at 22:14

1 Answer 1


Some unsupervised models use random functions and you might not have the same clusters as before.

Nevertheless, you can apply some functions to know the clusters features'ranges and define them with specific labels, so that you can identify future clusters easily (but not the ones out of the ranges, in that case you migh group them in a label "other" and reorganise them later).

  • $\begingroup$ Thanks Nicolas, I appreciate your contribution there $\endgroup$
    – EB3112
    Jul 30, 2021 at 11:09

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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