Distance metric as a (hyper-)parameter

[alexgcsa]

Hi,

I have been using X-Means here...Is it possible to use the distance metric as a parameter of the algorithm?

If not, this would be a nice parameter to have. Depending on data, the Euclidean distance is not appropriate. Looking at the code of X-Means, it seems to use only the Euclidean distance, not being possible to directly use the others, such as Chebyshev, Canberra, Gower or Minkowski.

Thanks in advance.

Alex de Sá

[annoviko]

Hi @alexgcsa ,

I have to check the current implementation and an article before I will introduce it just to be sure that there is no any contradictions.

[alexgcsa]

Thank you, @annoviko.

[annoviko]

Changes are available on 0.10.dev branch (in case of using sources from github you have to build C++ library by yourself - https://github.com/annoviko/pyclustering/wiki/Cpp-PyClustering) and will be available in the next release 0.10.1.

[alexgcsa]

Thank you again, @annoviko. I am going to try to use it this weekend.

Cheers,

Alex