[OSM-talk] [Talk-us] TIGER Addressing Import
aschoell at gmail.com
Mon Oct 5 05:22:20 BST 2009
On 3 Oct 2009, at 9:13 , Dave Hansen wrote:
> Is this the most up to date way of keeping addresses?
yes and well supported by the renderer
> Well, I have some perl code that will parse the 2007/2008 TIGER data
> files. My goal is to get the addresses imported into OSM this time
> around. Here's some ugly sample output of what I have now.
did't see any addresses there or any tags with anything useful. Is
this just a plain tiger conversion?
> I don't need any suggestions on it, I just wanted to show people
> what I
> have so far.
> * I currently have the way combination code turned off. This means
> OSM ways are represented 1:1 with the TIGER TLIDs.
please don't link any address creation code to use tlid attributes in
osm data. ways have been split, merged, moved, renamed
also if you base it on the latest 2009 data it may not align with the
imported tiger data.
this is a very difficult task and requires either manual upload and
corrections or an intelligent solution with roadmatcher or similar tools
to make it even worse you need to use deep history analysis to
guarantee that the data aligns
See this thread for details http://firstname.lastname@example.org/msg09008.html
> * Some multi-segment TIGER roads have only a single address range.
> Do we create segments for each road segment and evenly divide the
> addresses? Or, do we draw a single address segment going from the
> first to last node?
if a road is segmented because there is a crossing then it should be
split. if it's a single way on ground there is no need to split.
tagging for the renderer is bad. tagging for tiger roads will be evil
> And, yes, I saw the 2009 data come out. I'll make sure my code works
> with it.
please share the code when it's mature enough. I am sure many will be
interested in it and test for areas they know and can verify.
> -- Dave
> Talk-us mailing list
> Talk-us at openstreetmap.org
More information about the talk