Typing tool for Conjoint-based segmentation

Question

Hello all,

I am looking to conduct a segmentation using utilities derived from a conjoint exercise. When it comes to typing future respondents into segments, are there any best practices, suggested algorithms, papers you can recommend?

I am new to this and would appreciate any help.

Thank you.
-K

Answer 1

Hi, K, that's a really good question.  

We face a similar problem when building typing tools for MaxDiff-based segmentations and there we use a naive Bayes approach.  You can read more about this approach in the paper by Dave Lyon in our 2016 Conference Proceedings (which won the best paper award at that conference, no less).  

Better still for your needs might be the paper by Lech Komendant that starts on page 183 of those same Proceedings.  Lech descibes several ways of creating typing tools for segmentation based on choice experiments.  You might find the "pairwise" method described in that paper to be your best bet.