3
$\begingroup$

Can anyone please explain that is there any advantage of using hierarchical clustering over spectral clustering? I know how they work but I want to know in which situations its better to use hierarchical clustering over spectral clustering.

Thanks

$\endgroup$
  • $\begingroup$ Please don't post the same question twice (SO and here). $\endgroup$ – Anony-Mousse Apr 15 '16 at 20:54
2
$\begingroup$

Let's say that you know that there is a hierarchy in your data, and that you want to preserve this hierarchy. It will be easy to do that with hierarchical clustering but not necessarily with spectral clustering.

For example, spectral clustering with $K_2 > K_1$ will not necessarily yield a partition $P_2$ that is a refinement of $P_1$, i.e. some elements in a cluster of $P_2$ can belong to different clusters in $P_1$ which is not possible with hierarchical clustering since it yields a tree.

For concrete example, consider stock markets. They are run in different regions: Asia, US, Europe. Then, inside of each region, there are several industries, and then sub-industries. Assets behave (their comovements) according to this subdivision. In that case, it would make more sense to apply a hierarchical clustering and cut at the meaningful levels. A spectral clustering may not handle well this situation.

Well, there are also 'hierarchical spectral clustering'...

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.