Let's assume a crowd-sourced tree location use case where a research team asks hikers to GPS the location of a specific tree type in case they encounter it.

This will lead to a large number of longitude-latitude points each identifying the measured (noisy) GPS location of a tree. The researchers would now like to compute from those measurements a map of those trees.

That means they have to differentiate location point estimates for close trees and after assigning the estimates to a specific tree calculate the location of that tree.

Now trees don't move and GPS measurements can be pretty inacurate. What clustering algorithm would be the canoncial choice for this application?

(If you have an R package::function to suggest please tell me)

  • Did you search on this forum to see if k-means clustering might be useful? – user681 Jul 26 '14 at 12:50
  • Of course - k-means is the first you stumble upon and I also used it for similar but different use cases myself. Is K-means what a GIS pro would use for this task? – Raffael Jul 26 '14 at 12:58
  • that would require a value judgment and an endorsement :) but why not if your data meets the conditions, then give it a shot – user681 Jul 26 '14 at 13:26
  • of course I will - nonetheless I like to evaluate new options and this scenario seems to be a good situation as k-means is pretty generally applicable and I see potential for a more efficient algorithm given the specific situation – Raffael Jul 26 '14 at 13:30

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.