Hi Leif,
Thanks for your efforts on this! And Levente for your previous efforts
and now jumping back in for more!
I think we should include buildings in the import. Miami-Dade provides
560k buildings at
https://gis-mdc.opendata.arcgis.com/datasets/building-footprint-2d
In addition, I think we should mine the POI data available from the
county and parcel data if needed from the Property Appraiser. They
provide several data sets relevant to the cause:
Commercial Property - 88k records of points that is mostly businesses. I
did some investigation of this layer. This appears to be either from the
tangible tax file or closely related. Business names are included. NAICS
codes are there that can be cross-walked to OSM features like
amenity=restaurant. The records also include a folio (parcel
identification number) that allows joining to the site address data
(which has folio as well). And it includes addresses on its own. An
interesting issue is the point locations do not match the address points
layer. Appears to be the result of a different geocoder utilized to get
the point locations. And there are many times more commercial property
points associated with a parcel and its building(s) than address points
associated with the same parcel. Maybe discussion with Miami-Dade GIS
folks could shed some light on this. Bottom line is the address points
layer under-represents the business (POI) points that are represented
int he commercial property layer.
https://gis-mdc.opendata.arcgis.com/datasets/commercial-property
Gas Station - 699 records. Some point locations in this data set look
different than both addresses and commercial property and some match the
addresses layer. Includes the folio for joining to address data.
https://gis-mdc.opendata.arcgis.com/datasets/gas-station
Hotel Motel Inn - 593 rows - includes folio
https://gis-mdc.opendata.arcgis.com/datasets/hotel-motel-inn
More POI related layers from the county are available.
Parcels from the Property Appraiser - 911k records (including condos) -
folio, land use, site address, site city, site zip code, etc. Utilizing
land use codes would allow us to properly tag residential, condo, etc
buildings that are not represented in the commercial property layer. Can
grab missing zip codes and or city names from this data as well.
One other issue is some streets in Miami-Dade are referred to by two
names, and only one name exists in the address data. For example, locals
refer to Sunset Drive, but the official name in the address layer and
property appraiser data is Southwest 72nd Street. I have an older
alt_names.dbf, but need to take a closer look and see if an updated
table is available. It would be really nice to get the alt names in
there as well. This is an issue in other counties, too.
Brian
On 9/16/2018 3:47 PM, Leif Rasmussen wrote:
Hi everyone!
Me and some other contributos from the talk-us mailing list are
planning an import of every address point in Miami-Dade County,
Florida. The import will involve addresses from
https://gis-mdc.opendata.arcgis.com/datasets/eef6b33da60d47c0964387960c840eea_0.
They have already been converted to OSM XML format, and all duplicates
of existing addresses in the OSM database have been removed, leaving
only missing addresses in the dataset. The processed data is
available at
https://drive.google.com/open?id=1DJGNdONqdTXMlA0e550ghsmpotqc4QM4.
The import will be completed with the OSM US Tasking manager in the
next couple months. As of now, no addresses have been imported.
The important stuff:
* The import will add about 500,000 addresses in Miami-Dade County,
Florida over the course of several months.
* The addresses do not include any units. A seperate file is
available on the GIS site with the addresses that include units, but
it is 1.5 million points (3 times as many). The gain from having
"addr:unit" addresses is not as high as the cost of having to add,
maintain, and deal with them.
* The addresses don't have suffixes. Main Street East would have
"addr:street"="Main Street". This resulted from a transformation
error, and will most likely be fixed before uploading.
* The addresses have 5 digit postcodes, without 4 digit extensions.
The extensions are currently stored in a separate tag that can be used
or deleted later. If there is a good way to add the extension, it
will be added.
* The addresses are in the public domain. Licensing will not be an issue.
* No buildings, POIs, or other features will be imported - just
address points.
* The import will use an OSMUS Tasking Manager project to organize
uploading. The link to it project is located below.
* Conflation will not be a major issue. Since all duplicates of
existing addresses have been removed from the dataset, it is
technically possible to just import the addresses without any manual
checks. This would result in all missing addresses being neatly added
around the existing ones.
* Addresses will be given the tags "addr:housenumber", "addr:street",
"addr:postcode", "addr:city", and "addr:state". Source tags are
unnecessary, but if people want them, they can be added.
Links:
Original address source:
https://gis-mdc.opendata.arcgis.com/datasets/eef6b33da60d47c0964387960c840eea_0
Processed data for uploading:
https://drive.google.com/drive/u/1/folders/1DJGNdONqdTXMlA0e550ghsmpotqc4QM4
Import wiki page:
https://wiki.openstreetmap.org/wiki/Miami-Dade_County_Address_Import
Tasking manager project: https://tasks.openstreetmap.us/project/72
Please let me know if you have any concerns, suggestions, or things
you would like to see in this import. I would love feedback on
anything at all. :)
Thanks,
Leif Rasmussen
_______________________________________________
Imports-us mailing list
[email protected]
https://lists.openstreetmap.org/listinfo/imports-us
_______________________________________________
Imports-us mailing list
[email protected]
https://lists.openstreetmap.org/listinfo/imports-us
