Earlier, thanks for reading my articles. First, I wanna compare marketing analytics its called RFM analysis (marketers would be know this analysis) and the machine learning method for clustering is KMeans. In the final result we know RFM give 7 cluster, and Kmeans 4 cluster. Second, Who is the best?
In my humble opinion, when we build model clustering so many assumption that we have before do to the modeling or analysis. So, we can describe the objective first. In case you have a task for grouping 5 cluster from your data (in case the task from your company). And the best from your analysis (both of them) give 7 cluster. The conclusion is back to you, what if we have 7 cluster? is possible? or we back to our target to grouping on 5 cluster.
Maybe, it's an example simple case that I have built in some clustering method.