# HDBSCAN cluster: still unclear to me how to chose 'min_cluster_size

Hdbscan is an excellent technique to find the "optimal" number of clusters within your data when you have little a priori idea how many clusters should exist. This makes the method great for exploratory analysis:

http://hdbscan.readthedocs.io/en/latest/comparing_clustering_algorithms.html

Here's my problem: All results using hdbscan with the python implement in the link above rely on the crucial min_cluster_size

If users have a priori little idea how many clusters best fit the data, what is the correct approach above? Isn't there a metric one uses to decide what the optimal number of clusters is?

## 1 Answer

Optimal in which sense?

The crucial thing with clustering is that there is no optimal solution. Different solutions tell you a different part of the story. And to be able to get different views, you will need parameters. It is a exploratory technique.

Various attempts at defining "optimal" solutions have failed for practical use, just think of k-means.

• By "optimal", I mean certain clusters fit the data better than others. – ShanZhengYang Feb 7 '17 at 21:36
• Define "fit the data". min_cluster_size=1` supposedly is the "best fit" yet subjectively usually much worse. There exists no objective notion of "best" that people find useful... – Anony-Mousse Feb 7 '17 at 22:03
• So what's the point of clustering? – ShanZhengYang Feb 8 '17 at 17:09
• Tool to give you ideas how to look at your data. Nothing automatic, too unreliable. – Anony-Mousse Feb 8 '17 at 20:38
• We're off the beaten track, but after getting a set of clusters...what am I supposed to infer with these? Investigate the properties of clusters? The consequences of the techniques are slightly strange – ShanZhengYang Feb 9 '17 at 20:01