Closed mikelmaron closed 6 years ago
:wave: @ramyaragupathy and I are working on 2 things for this:
The current annual snapshots located at http://osmlab.github.io/osm-qa-tiles/historic.html were generated with the multipolygon
flag set for minjur
and contain duplicated data: multi-polygons are often represented as both multi-polygons and line strings, and sometimes even polygons. As such, you’ll notice the sizes of the tilesets varies (grows) drastically. As such, data is duplicated / triplicated.
Example: http://mapbox.github.io/osm-analysis-collab/user-summary?minEdits=1&minPercent=0&userList=[%22PlaneMad%22]&yearIdx=10.21/37.931/-87.070.
While PlaneMad
edited a lot in 2015, he didn’t actually edit a feature on every tile inside Indiana, but rather edited the relation that is Indiana, version 117, here: https://osmlab.github.io/osm-deep-history/#/relation/161816.
The multipolygon that is "Indiana" propagated the single edit to the administrative boundary to every tile contained within. If counting edits, objects, or users with these tilesets, they will be dramatically over-represented by this propagation.
https://gist.github.com/jenningsanderson/9350924d1a692e631ac72b09ef49a94b
2016-Q4 (2017-01-01) historic snapshot is available here: https://s3.amazonaws.com/mapbox/playground/ramyaragupathy/snapshots/2016-Q4-qa-tiles.mbtiles
@geohacker -- can we find a permanent place for these quarterly snapshots and update http://osmlab.github.io/osm-qa-tiles/historic.html accordingly?
TODOS:
The historic.html page has been updated: https://osmlab.github.io/osm-qa-tiles/historic.html
@ramyaragupathy @jenningsanderson