GeoNames Forum

I was wondering if anyone has attempted to set free some clustering algorithms on geonames-data so to group poi's together and to extract meaningful info.

For instance cluster monuments for getting a feel of the 'old area' in a city.
cluster hotels for business as well as touristic areas, etc.

Would there be any good reason why clustering on geonames-data in particular may give unuseful results?

Obviously, a lot of data is stil missing especially for these poi's as hotels, monuments, churches, etc. But I predict (hope) that when you cluster on a city level the missing poi's have about the same eucledian distribution as the already existing poi's , thus not changing the generated clusters based on the partial dataset very much. Would that be wishful thinking?

Moreover, for people who have already tried something like this, please share some info on what type of clustering algorithms you have used. As I'm not really an expert in the field I could use some guidance to get started.

Thanks,
Geert-Jan

Hi Geert-Jan

Tom Gibara has written a fast spatial clustering algorithm:
http://www.tomgibara.com/clustering/fast-spatial

Best

Marc