What distance measure does centroid based meanshift algorithm uses for clustering?

Question

I have a sparse TruncatedSVD matrices with 810 * 74 dimensions. I am trying to cluster with centroid-based MeanShift algorithm using sklearn api. However, I could not find information about the distance measure that meanshift uses by default neither in sklearn documentation nor in online blogs. There is no option to provide a precomputed distance matrices too. Does Meanshift caluculates Euclidean distance by default before doing centroid based computation?

Gone through sklearn documentation - https://scikit-learn.org/stable/modules/generated/sklearn.cluster.MeanShift.html

Also searched online but without a luck

score 0 · Answer 1 · answered Apr 26 '19 at 13:54

0

It's not using distances directly.

It's using the RBF kernel, which does however contain something akin to Euclidean distance inside.

I doubt this is a good choice for 74 dimensional data.

answered Apr 26 '19 at 13:54

Has QUIT--Anony-Mousse

76,138
12
138
194

Thank you for the response. Do you suggest any other algorithm for 74 dimensions? Data is sparse too. – Abdul Basith May 02 '19 at 07:58
Depends on your data, so I cannot give you recommendations. With this n and p, I'd say tough luck to get anything sound working. – Has QUIT--Anony-Mousse May 02 '19 at 18:37

What distance measure does centroid based meanshift algorithm uses for clustering?

1 Answers1