-1
$\begingroup$

Is there some other clustering algorithms apart from K-means in which I can define no of clusters I require ?I have a data set of large and skewed data points and K-Means is not providing quite satisfying clusters as in K-means two nearer point may end up with completely different clusters depending upon the cluster center points I start with.I want to try some other algorithms in which I can specify no of clusters I require(Not some hierarchical algorithms like mean shift).

$\endgroup$
  • $\begingroup$ Yes, there are plenty. Have a look at ELKI which probably the largest collection of clustering algorithms. $\endgroup$ – Anony-Mousse May 27 '16 at 12:00
1
$\begingroup$

A common approach with k-means is to run multiple times with different initializations, then pick the best. Some other options that let you pick the number of clusters: Gaussian mixture models, agglomerative clustering, various methods based on graph partitioning problems, other k-means-like algorithms (e.g. k-medoids), spectral clustering (which is really just k-means after a nonlinear embedding).

$\endgroup$

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.