Back to clustering algorithms: a look at a classic, DBSCAN (and its new H-version).

So you’ve tried PCA to hopefully look at clusters. You’ve scaled your data. You’ve tried both 2D and 3D. You’ve tried scaling your data in other ways. Thrown the Box-Cox at it.

The problem with Data Science these days is that nobody seems to agree on what it really is. Wikipedia generously calls it “interdisciplinary”; a quick search on LinkedIn reveals job scopes that vary from Analyst roles to Deep Learning.