8

Given a database of geographical locations (long/lat), what would be the best approach to determining/detecting clusters of locations that are within x miles of the cluster center AND total at least y locations?

e.g. out of 1000 McWidgets in NC, there are 30 clusters each containing 20 or more stores within 7 miles of their respective cluster center.

It's been a long time since my applied math course in college... any help for an old mushy brain would be greatly appreciated.

etriad
  • 459
  • 2
  • 6
  • 12

2 Answers2

6

A common method for this kind of problem is Density-based Spatial Clustering of Applications with Noise (DBSCAN). A variation which may be a better choice, if you can't determine a good density parameter, is the Ordering Points To Identify the Clustering Structure (OPTICS) algorithm, which uses a distance parameter, rather than a density parameter.

andand
  • 17,134
  • 11
  • 53
  • 79
1

You probably need one of the Clustering algorithms.

Tomas
  • 57,621
  • 49
  • 238
  • 373