I have two data sets:

  1. Customer demographic data;
  2. Transaction data of the customers.

Now, if I have to identify potential customers to develop a marketing strategy, I would make use of clustering to identify the similar groups and develop a marketing strategy accordingly.

How could I proceed to identify potential customers in a new list of customers not previously seem by the clustering algorithm? Is it like we apply the interpretations we observed from the clusters formed to the new customers??

Need some guidance on this.

  • $\begingroup$ It depends of how you performed your clustering, but scikit-learn and other libraries output a classifier for the clusters. You could also add the new data to the old one and reperform clustering. $\endgroup$ – Pedro Henrique Monforte Aug 3 at 18:46
  • $\begingroup$ @PedroHenriqueMonforte could you please give a brief about this point "how you performed clustering" .Im not clear on this point. $\endgroup$ – swetha Aug 4 at 6:53
  • $\begingroup$ Ofc. Just give us some sense of what you did or plan to do, for example which clustering method you chosen and which libraries/language your used for your code. $\endgroup$ – Pedro Henrique Monforte Aug 5 at 2:13
  • $\begingroup$ @PedroHenriqueMonforte Im planning to use agglomerative hierarchical clustering and scikit learn library for the same $\endgroup$ – swetha Aug 5 at 13:58

You are absolutely right. This method is called customer segmentation. Here we cluster customers based on many features like their demographic and income.

Suppose we get to know a particular set of people who belong to high income group/rich demographic are not spending much then we can do promotions to increase sales to get potential new customers hence it help drive business decision.

Having said that it's not best to run the algo once on past data and make decision as we know that customer behaviour changes frequently .So,companies usually run clustering algorithm daily to take advantage of the new customer behaviour.

So, to make our approach robust we need to find optimal no of clusters, choose best features hence it is a iterative process.

Refer this to know more.

| improve this answer | |
  • $\begingroup$ That's relevant information, but I think he just doesn't understand how to apply the results of the clustering to classify new data. $\endgroup$ – Pedro Henrique Monforte Aug 3 at 18:44
  • $\begingroup$ Yeah that's why i mentioned a tutorial link to clear the doubts. Also as mentioned it's best to rerun with new data. $\endgroup$ – prashant0598 Aug 3 at 18:49
  • $\begingroup$ thanks for sharing the link @prashant0598. $\endgroup$ – swetha Aug 4 at 6:50
  • $\begingroup$ You're welcome. Hope it answers your question. $\endgroup$ – prashant0598 Aug 4 at 6:55
  • 1
    $\begingroup$ hey @swetha it is not proper to say "thank you, you solved my problem" or "that is helpful" on comment section instead accept or upvote the answer (respectively) $\endgroup$ – Pedro Henrique Monforte Aug 5 at 2:15

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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