Resolved: Why to use PCA + KMeans for Purchase data?
I have understood that for segmentation part of the project, on applying only KMeans we were getting bad clustering. Hence we used PCA + KMeans in our segmentation data.
q1. But in purchase data if we only use KMeans, we are able to get Purchase prob, Brand choice, Purchase quantity, so why not skip PCA here?
1 answers ( 1 marked as helpful)
Hi Aditya,
thanks for reaching out and great to have you on the course. PCA is not a necessary prerequisite to k-means, so you can skip it if you'd like. Generally, PCA helps get better results on clustering data, however, if you're confident in the results of the k-means only you can implement the clustering step that way also.
Hope this helps!
Best,
365 Eli