Timeline for How does ST_ClusterWitihin and ST_ClusterDBSCAN scale?
Current License: CC BY-SA 4.0
10 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Dec 17, 2021 at 20:03 | answer | added | dr_jts | timeline score: 6 | |
| Dec 17, 2021 at 17:40 | comment | added | geozelot |
That's what I wanted to know. And that's the foremost issue: as with basically all ST functions, whenever they accept a unit parameter, that unit relates to the units of the underlying CRS - so the given 10000 are treated as degrees, thus for each point in that table the algorithm fetches and processes all 5M records! Obviously, no index will be used then. Changing that will help, at least with the speed - I have a more thorough answer half done on scaling DBSCAN, and why you will likely still be disappointed by the results, even with the translated eps distance. But no time.
|
|
| Dec 17, 2021 at 16:54 | history | edited | vaer-k | CC BY-SA 4.0 |
added 1182 characters in body
|
| Dec 17, 2021 at 16:53 | comment | added | vaer-k | @JGH thank you for the tip. I've done so | |
| Dec 17, 2021 at 16:52 | comment | added | vaer-k | @geozelot I'm not well-informed on GIS yet; I think you're asking for the SRID? In that case, it is 4326 | |
| Dec 17, 2021 at 14:58 | history | edited | JGH | CC BY-SA 4.0 |
improve code formating
|
| Dec 17, 2021 at 14:55 | comment | added | JGH |
The doc says to run analyze after a cluster command
|
|
| Dec 17, 2021 at 8:27 | comment | added | geozelot | Just to be sure: are those geometries projected (what CRS)? | |
| S Dec 16, 2021 at 21:57 | review | First questions | |||
| Dec 16, 2021 at 22:07 | |||||
| S Dec 16, 2021 at 21:57 | history | asked | vaer-k | CC BY-SA 4.0 |