Good Afternoon,
I’m currently planning a web-app/service project with a geolocation-enabled user model (lat/lng etc) and I was wondering what would be the best approach to find out the n biggest ‘hot spots’, e.g. geolocations with a given radius (e.g. 10 miles) where the most users are located at?
Does anyone know a good, practical clustering algorithm or other (existing) solution(s)? This is a pretty bird-view kind of question, I know… but backend technology wise I’m still open to anything as this particular feature is obviously only one of the whole feature set, but might help making a decision towards a particular set of tools/languages/environments.
Cheers & thanks,
-J
SQL Server’s spatial data types would be worth a look. It allows you to index on the geography column and do queries for distance. Not sure how easy it would be to group by radius, but at least having the geography data type and building indexes on it should help a lot with this type of problem.
Geography Methods Supported by Spatial Indexes