Good, detailed piece in DSC on motivation for Spectral Clustering. Not a method we used in the enterprise, never getting beyond K means and forest methods, but the case is made that better clustering methods can get you closer to the actual decision process. Getting closer to selectively using specific data resources to match and improve existing decision processes is valuable.
Thus the link to decision tree oriented methods. Also related to segmentation methods driven by clustering. Finally, mentioned in the article, how this can be used to the specific value of particular data assets. That was an area we looked at closely. Will examine possibilities. Ideas?
Subscribe to:
Post Comments (Atom)
No comments:
Post a Comment