[postgis-users] Can geocode() deal with NYC Hyphenated address numbers and other special types?

[Stephen Woodbridge]

I do not believe it is able to deal with  milepost, grid-style, or 
hyphenated house numbers other than the fact that it can pull off the 
first number and might use that to interpolate.

First understand that it only do linear interpolation between two number 
in a range. These complex numbers do not lead themselves to simple 
linear interpolation. I did an analysis of these address number in the 
past in Tiger data. Specifically, I compared the address ranges, to look 
at which components of the number were constant between the start and 
end of the range and which parts varied over the range.
It was not obvious in many cases which numbers to use for the 
interpolation and in some cases it was totally impossible to interpolate 
because the numbers collected did not make sense (block number at start 
and single number at end) either through a typographical error in the 
data or based on who collected the numbers at the end of the segments or 
how numbers were reassigned when a segment got divided into multiple 
segments.


Also remember you are working with Tiger data which fuzzes the address 
ranges because of Title 13 concerns.

I'll let others respond to the specifics of the postgis geocoder.

-Steve

On 11/30/2015 3:56 PM, draco doc wrote:
> Hi,
>
> I found there are many special format of addresses that valid in their
> local area, but could be quite difficult for geocode() in Tiger Geocoder.
>
> Here is a US street address standard document
> <https://www.fgdc.gov/standards/projects/FGDC-standards-projects/street-address/index_html>.
> Started from page 36 of the final draft, several special cases were listed:
>
> 4. Milepost Complete Address Numbers (Example: "Milepost
> 240").
>
> 5. Grid-style Complete Address Numbers (Example:
> "N89W16758"). In certain communities in and around southern
> Wisconsin, Complete Address Numbers include a map grid cell
> reference preceding the Address Number.
>
> 6. Hyphenated Complete Address Numbers (Example: "5-
> 5415"). In some areas (notably certain parts of New York City,
> southern California, and Hawaii), Complete Address Numbers
> often include hyphens.
>
> I have 18 million address to geocode, and all these special cases can be
> found in my data. From my test, geocode() seemed not be able to
> recognize these cases which is totally understandable for me.
>
> My questions is, is it possible to modify input a little bit to get a
> better than nothing result?
>
> For example, if the NYC hyphenated address cannot be recognized, can we
> at least remove some part and get a rough location? I have address like
> this "40-12 28 AVE , New York City, NY 11103", where 40 is block, 12 is
> house number in block. I tried to just remove the 12 and keep the 40
> which is block number, but all I can get is still just the 28 ave.
>
> Thanks!
>
>
> _______________________________________________
> postgis-users mailing list
> postgis-users at lists.osgeo.org
> http://lists.osgeo.org/mailman/listinfo/postgis-users
>