So I am trying to cluster a dataset that looks like the following:

My Data

I have tried K-Means and GMM, which give me horrible results. I have tried DBSCAN, which was okay, but it is difficult to choose the parameters.

What techniques would you use? I am thinking about a Box-Cox transform, then using K-Means again, or trying a different distance measure (p<1) but I wanted to get some expert opinions as well.

So, what should I use?

  • $\begingroup$ Have you tried scaling the data? $\endgroup$ – honeybadger May 20 at 0:13
  • $\begingroup$ @honeybadger Yes I have. It just gives me smaller values on the axes but the problem still remains. $\endgroup$ – The Dude May 20 at 13:51
  • $\begingroup$ hey.. are you trying to cluster only using dimension 1 and dimension 2? First of all, dimension doesn't look like a continuous variable.. is it something else? $\endgroup$ – StupidWolf May 24 at 11:59

Your Answer

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

Browse other questions tagged or ask your own question.