[postgis-users] ST_ClusterDBSCAN for "geography" data type?

Darafei "Komяpa" Praliaskouski me at komzpa.net
Mon Dec 21 03:24:49 PST 2020


My last exercise in KMeans showed that it's enough to add support for 3D
distances instead of 2D distances in the code and transform your geometry
into EPSG:4978 (after Force3D). That will cluster in a 3D XYZ coordinate
system using straight lines in 3D, which is usually good enough.

On Mon, Dec 21, 2020 at 2:15 PM Giuseppe Broccolo <g.broccolo.7 at gmail.com>

> Hi Marco,
> Il giorno dom 20 dic 2020 alle ore 10:03 Marco Boeringa <
> marco at boeringa.demon.nl> ha scritto:
>> Hi,
>> Reading through the PostGIS documentation, I noticed the
>> "ST_ClusterDBSCAN" function takes a distance as one of the inputs. Now
>> the docs suggest the current algorithm only takes in "geometry" type
>> data. Is that true? Based on the distance input variable, it would seem
>> logical to have a "geography" variant as well, even if that is a
>> considerably slower variant considering the more complex distance
>> calculation.
> Yes, ST_ClusterDBSCAN takes just the geometry type in input. This is
> because most of the algorithm
> is implemented using utilities already existing in GEOS for planar
> geometries.
> Adding the support for the geography type would mean to re-implement the
> algorithm specifically for
> spherical objects, as you mentioned.
> Regards,
> Giuseppe.
> _______________________________________________
> postgis-users mailing list
> postgis-users at lists.osgeo.org
> https://lists.osgeo.org/mailman/listinfo/postgis-users

Darafei "Komяpa" Praliaskouski
OSM BY Team - http://openstreetmap.by/
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.osgeo.org/pipermail/postgis-users/attachments/20201221/9c3822e7/attachment.html>

More information about the postgis-users mailing list