Solved

Boost NeighborFinder, is that possible?


Badge +11

Hi all,

I'm having a bottle-neck on a workbench due to the use of the transformer "NeighborFinder". This operation only is taking around 18 hours.

 

The goal is to find around 200K points to the nearest Polygon (having around 10K). Points that I want are quite near, not touching some times. Also, there are some points that are farther and I don't want to have them on the "MatchedBase" output (red ones):

sample 

These are the transformer parameters:

 

 

nf 

 

Also, attributes from Polygon need to be transferred.

I don't see the option to make the search by Groups, since I cannot foresee which points belongs to certain Polygons beforehand. Also, reducing the number of features is not possible.

 

Any suggestion?

Something that could be faster on that search?

Thanks!

Juanma,

icon

Best answer by redgeographics 11 March 2022, 11:55

View original

3 replies

Userlevel 4
Badge +25

Just throwing a random idea out there: what if instead of using the NeighborFinder you buffer the polygon slightly and then use a Clipper? It might be faster.

Badge +11

Hi @Hans van der Maarel​ ,

I was about to reply to myself with that solution. Thanks!

What I've done is to switch that NeighborFinder with a Buffer on Polygons (same amount as NFinder) and PointOnArea, after that a simple Tester to know what's inside:

 

imageIt is now taking just a few minutes!

Thanks for coming with that idea too, @Hans van der Maarel​ !

Juanma,

 

Userlevel 1
Badge +11

Hi @juanmahere​ Can you share your data and workspaces with us? It looks like we need to get the NeighborFinder to catch up to the PointOnAreaOverlayer! https://community.safe.com/s/submit-case Thanks!

Reply