I am looking for a process in ArcGIS that takes a point feature class and groups the points into a user-provided number of clusters.

  • 1
    Have you explored the questions associated with the clustering tag?
    – whuber
    Jan 6, 2012 at 20:38
  • @whuber I poked around for a while on the stack exchange without much luck. I'll explore and let you know what I find
    – Kjohnson
    Jan 9, 2012 at 23:11
  • I found a bunch of good answers here. Perhaps more to the point, a brief survey of the variety of cluster-related questions will suggest the additional information you should be supplying to get a good solution to your problem.
    – whuber
    Jan 9, 2012 at 23:39
  • I didn't find anything directly corresponding to what I'm looking for. However, Clusterpy has provided me with the impetus to just go ahead and try and script a solution in Python anyway. I find it strange that there's no obvious tool/solution for this in ArcGIS.
    – Kjohnson
    Jan 10, 2012 at 21:45

2 Answers 2


k-means clustering is exactly what you want. Version 10.1 has k-means clustering... if you are beta program eligible you could try that. You could also try doing your k-means clustering in R and then bringing the results over the arcgis. I think there are also some 3rd party scripts out there for k-means clustering using versions 9.3 and 10.0 too.

  • thanks for the suggestions. I'll snoop around and see what 3rd party scripts I can find.
    – Kjohnson
    Jan 9, 2012 at 23:08
  • Have a look at Geospatial Modelling Environment (GME) - spatialecology.com/gme/kmeans.htm It requires ArcGIS 10.X and R. Oct 16, 2015 at 16:21

If your data is attributed with some sort of "by type" field, you could use the Mean Center geoprocessing tool and set the case field.

  • @ Hornbydd, good suggestion. However, I'm looking for a method to group points based on geographical proximity only. Thanks though.
    – Kjohnson
    Jan 9, 2012 at 23:42
  • It is very useful this geoprocessing tool.
    – juasmilla
    Oct 3, 2014 at 9:27

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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