1
$\begingroup$

I'm looking for the best way to check if our customers have the correct segmentation label.

(New) customers are given a segment label at creation, mainly based on the information available at that time. There is a fixed list of segments.

We would like to check if there are customers for whom the segment might not be correct, based on their buying behavior and other parameters. We can't really use clustering (unsupervised), since we already have a fixed list of segments (which will not change). Anyone have an idea on how we should approach this question?

Thanks!

$\endgroup$

1 Answer 1

1
$\begingroup$

There are several ways to use distance between a customer and a segment represented by a group of customers in this segment.

One very simple way would be "train" a k-NN classifier to predict the segment a customer belongs to. Actually k-NN no training, the prediction for any new instance is based finding which instances in the training set this instance is the closest to. So if a customer you want to test is predicted a different segment as the one which it was assigned, it's potentially the incorrect segment.

$\endgroup$

Your Answer

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

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