osmlab / dcbuildings Goto Github PK
View Code? Open in Web Editor NEWRestart that ole DC building footprint + address import
Restart that ole DC building footprint + address import
This is a sidequest as this project is about buildings and addresses, but while we're at it:
Compare OSM streets with latest streets data available from OCTO DC.
What's the right OCTO streets file to use here?
For complex buildings like http://www.openstreetmap.org/browse/relation/1029369 the OSM tag data is saved in the relation. Currently, convert.py does not handle multipolygons; we'll need to devise a way to do that (probably via differences) and programatically create a relation to append the tag data to. This will also be a problem in reverse (managing multipolygons in the planet.osm file to see whether the building already exists in the OSM data).
From @pnorman:
The multipolygon generation is incorrect. See 1664 and 1666 at lat=38.9122
lon=-77.0612
We're currently just using rtree and a bounding box test to check whether an address location is within a building.
This is likely too coarse, leading to many false positives.
@lxbarth Was the possibility of regular updates to DC OCTO or other interested parties using changewithin ever discussed, similar to what you're doing in NYC? Is there a license impediment?
@iandees - How does one delete a task? Can you do this?
http://tasks.openstreetmap.us/job/5 is obsolete.
Running validation on JOSM results in "Duplicated nodes" errors.
The DC building dataset has many instances where two points sit in the exactly same location.
Generate .osm files chunked by census tracts (or similar). These files can then be loaded up in JOSM, reviewed and copied bit by bit into OpenStreetMap. They should be as clean as possible (well formatted tags, address nodes where more than one address per building) and not contain any existing buildings.
So at the last MappingDC happy hour a guy who works for the new DC GIO Tim Abdella was there and said he was sent to see how their office could help contribute to OSM. I figured I'd relay the message in case you guys needed anything else from DC as it looks like you're pushing forward with the import.
Do a diff between the official DC building dataset and the current building data in DC. Just two layers overlayed with each should do the trick.
Like we did for NYC https://github.com/osmlab/nycbuildings/blob/master/PROPOSAL.md
Like here osmlab/nycbuildings#8
The script currently only populates addr:housenumber
and addr:street
GIS_ID
-> dcgis:gis_id
? Previous imports included this.addr:city
and ZIPCODE
-> addr:postcode
? Previous imports did not include this.dcgis:captureyear
, dcgis:lot
, dcgis:square
, source=dcgis
, dataset=buildings
Example: building attributes from a previous import
Example: address attributes from a previous import
There's no reason to export separate building and address.osm files. Merge them into one building.osm file containing buildings with addresses and separate address points.
@ajashton reported:
There are some odd multipolygons being created for some of the rowhouses:
The courtyardish area is incorrectly being added as a second 'outer' polygon - it should not be included at all, and there shouldn't be a need for a multipolygon.
Needs further investigation. Here's what I suspect:
convert.js
uses intersects()
to test for a match between a void polygon and a building, delivering a positive on a polygon on such void spaces. (I couldn't get within()
to work properly)I noticed in a dataset I've been experimenting with that census block group boundaries can intersect building geometry, so that a simple intersection check will generate duplicate buildings at those boundaries, as those cases intersect multiple block groups.
I dealt with this by assigning buildings to a block group via contains, which does not report intersecting geometries. I then used an intersection checks; which assigns all buildings not picked up by contains to a single block group.
@lxbarth had a poorly fleshed-out thought on the way to work this morning.
The address data contains a field for the census tract that each address lies inside of. Using psycopg, we could potentially
This way we only have to read 1 census tract's worth of address into memory to perform the spatial join against the building data set.
Thoughts?
Compare addresses in OSM to latest address file OCTO DC. Compile a list of addresses missing in OSM ready to be used for mapping in OSM XML format.
What's the right OCTO address file to use here?
WIth multiple individuals reviewing and uploading data, should we:
Pardon me if this was decidedly not included in the import. It seems that many of the data are missing the addr:city
key, as well as addr:state=DC
debatable and addr:country=US
. To observe these issues, either use Unvollständige Adressen or run this query on Overpass Turbo:
<osm-script output="json" timeout="25">
<!-- gather results -->
<union>
<query type="node">
<has-kv k="addr:housenumber"/>
<has-kv k="addr:city" modv="not" regv="."/>
<bbox-query {{bbox}}/>
</query>
<query type="way">
<has-kv k="addr:housenumber"/>
<has-kv k="addr:city" modv="not" regv="."/>
<bbox-query {{bbox}}/>
</query>
<query type="relation">
<has-kv k="addr:housenumber"/>
<has-kv k="addr:city" modv="not" regv="."/>
<bbox-query {{bbox}}/>
</query>
</union>
<!-- print results -->
<print mode="body"/>
<recurse type="down"/>
<print mode="skeleton" order="quadtile"/>
</osm-script>
Use census block groups for more granular chunks to upload:
/cc @mikelmaron
This is what we're using now to chunk up data.
Update: edited to use block groups.
This was discussed and agreed on in the imports-us Google hangout on Monday Aug 5.
Create a light HTML web site guiding the upload workflow
Since the shapefile for the addresses includes information about the parts of the street name, consider including that in the OSM data too.
I did this in Chicago. Not sure if it's particularly useful yet, but I figure why lose some information about the street name by only including the merged name.
Expand all NW
and Rd
etc. Addresses should match street names.
A declarative, efficient, and flexible JavaScript library for building user interfaces.
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
An Open Source Machine Learning Framework for Everyone
The Web framework for perfectionists with deadlines.
A PHP framework for web artisans
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
Some thing interesting about web. New door for the world.
A server is a program made to process requests and deliver data to clients.
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
Some thing interesting about visualization, use data art
Some thing interesting about game, make everyone happy.
We are working to build community through open source technology. NB: members must have two-factor auth.
Open source projects and samples from Microsoft.
Google ❤️ Open Source for everyone.
Alibaba Open Source for everyone
Data-Driven Documents codes.
China tencent open source team.