4

I have the following 9 clusters (each colour represent the appropriate cluster):

enter image description here

I am looking for a clustering algorithm (if any) that would achieve "realistic" results.

The data represents Vancouver's bus stops, crossings, gates, stations etc (as you can see in the left side of the image).

I 've tried K-Means and I obviously had some bad results since it uses the Euclidean distance as its metric and that's not applicable in this scenario. I am using WEKA.

PolyGeo
  • 65,136
  • 29
  • 109
  • 338
Menelaos Kotsollaris
  • 335
  • 3
  • 9

1 Answers1

5

The method and script described in Graph/Network building and analysis of linked polygons in ArcGIS for Desktop? are applicable to any network, e.g.

enter image description here

Community
  • 1
FelixIP
  • 22,922
  • 3
  • 29
  • 61
  • Thank you for your answer. Do you have the referenced algorithm's description and/or pseudocode? It would be much clearer to me to understand what you are proposing. – Menelaos Kotsollaris Nov 16 '15 at 02:21
  • 1
    Pseudocode description http://gis.stackexchange.com/questions/123289/grouping-village-points-based-on-distance-and-population-size/123297#123297 – FelixIP Nov 16 '15 at 02:26