I was asked the following question during an interview:
What method would you use to segment customers into high, medium and low value groups based on their purchase history of last one year?
I said using RFM(recency, frequency, monetary) method. The interviewer seemed unsatisfied with this answer.
Can I use cluser analysis? If so, how to do it? I can plug the data in and ask the program to produce 3 clusters. But these 3 clusters may not have anything to do with customer values.
Any other method? I think this is an unsupervised learning problem.