Skip to main content
Solved

Networkcostcalutor: I'm doing something wrong with the network preparation


michiedem
Contributor
Forum|alt.badge.img+7

I am trying some network analisys (isochrones and route/gpx creation) and I am obviously doing something wrong, I just can't figure out what.

I am using OSM highways, unfiltered (I also tried a local route network, same result). The only manipulation I did is convert it to espg 28992RDnew (because it's in meters) before using it as an input for the topologybuilder (also tried topology calculator).

 

Since I have multiple start points I created a road network Per stop_id and used ‘group by’ on the topologybuilder and networkcostcalculator.

 

 

The results are very weird. Connected and (way too many) not connected seem random:

 

(green = connected, red = not connected)

 

Zoomed in:

I tried pretty much all settings on the networkcostacalculator.

 

When I try the shortestpathfinder it's also clear that something is off, it refuses to use some (random) paths/roads/lines. So it's obvious I am doing something wrong, I just cannot figure out what.

 

 

Best answer by nielsgerrits

Not sure what is happening. When I use your workspace I can’t get it to work. When I build something from scratch no issues. Maybe something wrong in 2023.0.0? I used 2023.2 (and never ever use .0 versions for production work honestly)

I think it also has to do with you source points being that far off the vertices in the network. For it to work it needs to be snapped to a vertex, not a segment. If I set Snap Source Points in the NetworkCostCalculator to 100 meters it does work for me.

Also, one of the source points in your sample set has the x / y swapped as it lands in Germany.

===

Looking through the Changelog:

NetworkCostCalculator: Inconsistent results when source node is not a valid node in the network
https://fme.safe.com/changelog/?_change_log_platform=fme-form

View original
Did this help you find an answer to your question?

8 replies

michiedem
Contributor
Forum|alt.badge.img+7
  • Author
  • Contributor
  • June 18, 2024

I just noticed sometyhing very weird: If I remove all attributes from the input point except strops_id (and nothing else), it changes the amount of roads that are connected. That shouldn't happen??


liamfez
Influencer
Forum|alt.badge.img+34
  • Influencer
  • June 18, 2024

Have you tried using the TopologyBuilder without a Group By?

A LineCombiner may also be helpful to combine some of those lines between intersections, where appropriate.


nielsgerrits
VIP
Forum|alt.badge.img+53

Hard to say what is going on, but I would start with one stop input. Put a Sampler before the first counter set to 1st N and get it to work, then process more or all features.


michiedem
Contributor
Forum|alt.badge.img+7
  • Author
  • Contributor
  • June 18, 2024
nielsgerrits wrote:

Hard to say what is going on, but I would start with one stop input. Put a Sampler before the first counter set to 1st N and get it to work, then process more or all features.

It works as intended with one stop. It goes wrind as soon as I add another one. Something else very weirs: I use the featurereader to read the OSM lines which I stored in postgis. I use an area (buffer around the point) with the stop_id. So all roads in that area get an extra attribute ‘styop_id’ which is the id of the stop. So far, soi good.

 

When I use the line combiner and combine on the stop ID I should get 2 sets of lines if I use 2 stops as input. If I open the attribute table there are only 2 sets of ID's. But the line combiner gives me a couple of hundred lines.

 

It must be something small/stupid, but I just can't find it


michiedem
Contributor
Forum|alt.badge.img+7
  • Author
  • Contributor
  • June 18, 2024
liamfez wrote:

Have you tried using the TopologyBuilder without a Group By?

A LineCombiner may also be helpful to combine some of those lines between intersections, where appropriate.

Same random result unfortunately :(

 

I also tried a different source for the network (for the Dutch people: I also tried the NWB). Same random result as soon as I add a second stop


michiedem
Contributor
Forum|alt.badge.img+7
  • Author
  • Contributor
  • June 18, 2024

Here is the workspace with some sample data. The result on my computer:

 


nielsgerrits
VIP
Forum|alt.badge.img+53
  • Best Answer
  • June 18, 2024

Not sure what is happening. When I use your workspace I can’t get it to work. When I build something from scratch no issues. Maybe something wrong in 2023.0.0? I used 2023.2 (and never ever use .0 versions for production work honestly)

I think it also has to do with you source points being that far off the vertices in the network. For it to work it needs to be snapped to a vertex, not a segment. If I set Snap Source Points in the NetworkCostCalculator to 100 meters it does work for me.

Also, one of the source points in your sample set has the x / y swapped as it lands in Germany.

===

Looking through the Changelog:

NetworkCostCalculator: Inconsistent results when source node is not a valid node in the network
https://fme.safe.com/changelog/?_change_log_platform=fme-form


michiedem
Contributor
Forum|alt.badge.img+7
  • Author
  • Contributor
  • June 21, 2024
nielsgerrits wrote:

Not sure what is happening. When I use your workspace I can’t get it to work. When I build something from scratch no issues. Maybe something wrong in 2023.0.0? I used 2023.2 (and never ever use .0 versions for production work honestly)

I think it also has to do with you source points being that far off the vertices in the network. For it to work it needs to be snapped to a vertex, not a segment. If I set Snap Source Points in the NetworkCostCalculator to 100 meters it does work for me.

Also, one of the source points in your sample set has the x / y swapped as it lands in Germany.

===

Looking through the Changelog:

NetworkCostCalculator: Inconsistent results when source node is not a valid node in the network
https://fme.safe.com/changelog/?_change_log_platform=fme-form

Just updating to the latest version did the trick. Thank you!!

 

(now I have a weird bug with the anchorsnapper though, I can only connect it to 1 input, whatever I do. Worked around it)


Cookie policy

We use cookies to enhance and personalize your experience. If you accept you agree to our full cookie policy. Learn more about our cookies.

 
Cookie settings