# Consensus methods for interpreting results of a cluster analysis

With cluster analysis, sometimes we want to assign meaning to the nature of the identified clusters. This is subjective and often done by a single person (or very few people) with perhaps some additional post hoc interpretion by a collaborator or reviewer, etc. Surely it is better to share the results from a cluster analysis with a larger group of experts and then elicit their collective interpretation of results.

My question is: What methods exist to arrive at a consensus from a group of experts regarding their interpretion of a cluster analysis?

For example, I have been able to find published uses of cluster analysis to analyze data from a consensus analysis, but not the other way around!

My question has some relation to, but is distinct from, the challenge of assigning priors to the parameters of a distribution based cluster method.

Any suggestions regarding terminology to use in my search would also be helpful.

• assign meaning to the nature of the identified clusters. This is subjective Why subjective? In what meaning "subjective"? We cluster by some features. Further, there is also external features that could be found to discriminate among the clusters. We borrow meaning from these and those. Maybe you aim to ask about a different matter? – ttnphns Mar 13 '15 at 15:28
• 2 people can and do "borrow meaning" in different ways, so I would label this subjective. – D L Dahly Mar 13 '15 at 15:36
• Then I presume it is not a statistical analysis question. – ttnphns Mar 13 '15 at 15:45
• I am suprised by that comment, and couldnt disagree more. – D L Dahly Mar 13 '15 at 15:53
• I meant to say that this question, if statistic, is not a clustering question; it is the question of the analysis of expert opinion. – ttnphns Mar 13 '15 at 16:45