After clustering my data into k groups, I would like to determine for each of the clusters, which dimensions(variables) significantly describe that particular cluster. For example, lets say cluster A contains a majority of people from one age group purchasing a vehicle from particular segment. Cluster B contains people belonging to various age groups but majorly from one or two regions (out of 10-15 regions).

Hypothesis: Using entropy, determine for each cluster, which variables are homogeneous and use those variables to create a definition for that cluster by computing the mode (or some other descriptive statistic). For example,

Cluster A: Customers of 30-40yrs age group, purchasing vehicles from segment C1, etc

Cluster B: Customers from region R1, earning 30K-40K per annum, etc

Are there any pitfalls of using this technique?



Your Answer

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

Browse other questions tagged or ask your own question.