Author Topic: Project Statistics  (Read 12161 times)

0 Members and 1 Guest are viewing this topic.

Offline yakra

  • TM Collaborator
  • Hero Member
  • *****
  • Posts: 1721
  • Last Login:Yesterday at 11:23:56 pm
Re: Project Statistics
« Reply #105 on: November 07, 2018, 05:09:43 pm »
That's what I get for just doing a quick Wiki walk -- Maughan Library -> Strand Campus -> OK, that's where it is? Sure! :)
"LB" means London Borough, I take it?

A little more digging (No further programming yet, just some spreadsheetery) reveals:
This node is 13 steps away from the master node that represents the whole world.
It is the only node of this size.
How many points/unique locations it or its four children contain is still an open question.

How about nodes 12 steps away from the master node?
There are 20 of those, with the majority in the UK. Three are in Puerto Rico. One at the Cincinnati OH / Covington KY line. Seems natural, with dense road networks in these places. And there's one in... Portland, meaning I was probably a bit too generous with waypoints when designing US1Alt, ME77 & pals back in the day. :P

How about:
The least dense node with 1 or more waypoints?
« Last Edit: November 07, 2018, 05:24:15 pm by yakra »

Offline si404

  • TM Collaborator
  • Hero Member
  • *****
  • Posts: 724
  • Last Login:Today at 03:04:47 am
Re: Project Statistics
« Reply #106 on: November 07, 2018, 07:15:54 pm »
"LB" means London Borough, I take it?
Yep.
Quote
How many points/unique locations it or its four children contain is still an open question.
For it: 51 points (non hidden, all but 7 junctions between two routes). 69 edges between them (one a concurrency) and 26 between them and points not inside the box (two a concurrency).

I copyied (roughly in terms of actual locations) the HDX across onto a screengrab of the wpteditor with the box on


Offline yakra

  • TM Collaborator
  • Hero Member
  • *****
  • Posts: 1721
  • Last Login:Yesterday at 11:23:56 pm
Re: Project Statistics
« Reply #107 on: November 07, 2018, 07:18:27 pm »
Looks like we had almost the same idea. :D
« Last Edit: November 07, 2018, 07:31:58 pm by yakra »

Offline yakra

  • TM Collaborator
  • Hero Member
  • *****
  • Posts: 1721
  • Last Login:Yesterday at 11:23:56 pm
Re: Project Statistics
« Reply #108 on: November 07, 2018, 08:03:34 pm »
Here are the other 13th-level nodes, with a quick-n-dirty name I picked for each.

CodeQuick-n-Dirty Name
MAADCDCCCBDAAHosford-Abernethy
MABDDBADBBDACCarrick Hill
MABDDBBCADDCDBelfast-IMN Ferry
MABDDBBDCCDDCMoss Side
MABDDBCBBBACAHarold’s Cross
MABDDBDABBBABPoulton-Liscard-Egremont
MABDDBDABBBBAKensington
MABDDBDBCBCADSmethwick
MABDDBDBCBCBCDartmouth Circus
MABDDBDDBDBCDKensal Green - Harrow Road
MABDDBDDBDBDCEuston Square
MABDDBDDBDBDDMile End
MABDDBDDBDDBANine Elms Square
MABDDBDDBDDBBNew Cross Gate
MABDDDBDABAAASt Saviour, Jersey
MADAABBBCCCCAPortland ME
MADAACABBBDDACovington-Newport
MADCBACACDDBDJardines
MADCBACADCCAABuena Vista Santurce
MADCBACADCCACBosque Urbana del Neuvo Milenio

Go to yakra.teresco.org/tests/tmg/Quadtree/WaypointQuadtree.tmg.html
Zoom in a bit, search for the code in the vertex/edge list, and click.
Center your mouse, Alt+Tab away, move mouse away from canvas, Alt+Tab back*, and click the OSM or Google links.

*Yes, I plan to eventually improve that...

OKAY I AM GOING TO CLOSE DOWN MY WEB BROWSER AND STEP AWAY FROM THE COMPUTER NOW

Offline yakra

  • TM Collaborator
  • Hero Member
  • *****
  • Posts: 1721
  • Last Login:Yesterday at 11:23:56 pm
Re: Project Statistics
« Reply #109 on: November 08, 2018, 01:56:17 am »
https://github.com/TravelMapping/DataProcessing/pull/126/files
The WaypointQuadtree has been initialized all these years with the latitude and longitude bounds swapped. It hasn't broken anything, but just led to a sub-optimal distribution of child nodes.
After fixing it, we're down from 42757 to 35165 total nodes with the latest HighwayData.
Compare these side-by-side in two browser tabs:
http://yakra.teresco.org/tests/tmg/Quadtree/WaypointQuadtree-good-NoMerc.tmg.html
http://yakra.teresco.org/tests/tmg/Quadtree/WaypointQuadtree-bug-NoMerc.tmg.html

With the bug, The nodes along the western edge of the map have their eastern boundaries, and centers, asymptotically approaching -90°, rather than the normal -180°.
The de facto W bound of each node is -90°, but points W of there can still be inserted normally, because they're W of the parent node's center -- be it 0°, -45°, -67.5°, -78.75°, etc.
As more points are added and these nodes are refined, they effectively become narrower and narrower "stripes" stretching all the way west.

The mirror image applies along the eastern edge, of course.

A couple of our final unrefined nodes:
Code: [Select]
NE_A http://www.openstreetmap.org/?lat=35.191956&lon=-89.998627
SE http://www.openstreetmap.org/?lat=35.189209&lon=-89.998627
SW http://www.openstreetmap.org/?lat=35.189209&lon=-90.000000
NW http://www.openstreetmap.org/?lat=35.191956&lon=-90.000000
NE_B http://www.openstreetmap.org/?lat=35.191956&lon=-89.998627
...
SE_A http://www.openstreetmap.org/?lat=35.131531&lon=-89.998627
SW http://www.openstreetmap.org/?lat=35.131531&lon=-90.000000
NW http://www.openstreetmap.org/?lat=35.134277&lon=-90.000000
NE http://www.openstreetmap.org/?lat=35.134277&lon=-89.998627
SE_B http://www.openstreetmap.org/?lat=35.131531&lon=-89.998627
A bit smaller than central London, innit! ;)

With latitude, things are less interesting: there are just four extra nodes each N of the N pole & S of the S pole, that never receive any waypoints.
« Last Edit: November 08, 2018, 11:20:55 am by yakra »