I think I understand :) On Fri, Oct 14, 2011 at 18:16, Ben Abelshausen <ben.abelshau...@gmail.com> wrote: > ...If no street with > the same name can be found the closest street is taken. > > This all returns a distance and then we filter all these that have distances > higher than let's say 50m. This gives us a good estimate of where roads are > missing! >
This'll help avoid spelling mismatches, but doesn't it mean that a missing address won't show up in the heat map if it's within 50m of any other street, even the wrong one? Any idea if this is a significant problem? Say, for example, addresses on a city alley missing from OSM but near another street? > I can share lat/lon/distance records that we use to generate the 'heat'maps. > I think there are about 50000 with a distance > 50m. The records will have > no address data (postalcode, street, housenumber), that is the part i am not > supposed to share. So, a table of lat/lon/dist values where you know of an address but no street is nearby, from which we can generate heatmaps. It should be small enough to throw up on a website somewhere, and then we just need a script to do the rendering. I'd be interested in working that out (probably possible with GDAL or similar) but I'm away next week so it won't be for right away. Cheers, Jw _______________________________________________ Talk-be mailing list Talk-be@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-be
