[Talk-GB] Non-intuitive addresses

[Mark Goodge]

On 15/02/2022 14:39, ndrw6 wrote:
> On 14/02/2022 13:25, Mark Goodge wrote:
> 
>> There is a complicated and roundabout way to infer from open data 
>> sources that these addresses are probably part of Huxleys Way. That 
>> is, take the coordinates of the street from the Open USRN database, 
>> match those up with OS OpenRoads, then look up the name in OS 
>> OpenNames (which gives us Huxleys Way), then find all the UPRNs which 
>> are associated with that USRN in OS Open Linked Identifiers, check 
>> their coordinates in OS Open UPRN and note that this includes a set of 
>> houses that, visually, appear to be on Boat Lane. But that's not 100% 
>> reliable, and there's no simple way of doing that with any of the OSM 
>> editing tools, at least as far as I'm aware.
> 
> Looks like a potentially useful tool to implement. There is a good 
> chance any of these steps would fail but any extra bit of information we 
> could infer from open data would be useful.

I've done something like that on https://uprn.uk. It works for Huxleys Way:

https://uprn.uk/usrn/45400706

But, unfortunately, there are cases where the UPRN-USRN association 
doesn't necessarily follow the postal address. For example, here in Ely, 
Cambridgeshire:

https://uprn.uk/usrn/12000472

As well as the properties that do have Broad Street as their address, it 
also includes properties with addresses in several other streets to the 
south-east of Broad Street, including Cardinals Terrace. But, oddly, 
Cardinals Terrace does also have a different set of UPRNs that are 
associated with it:

https://uprn.uk/usrn/12003034

I have no idea why there seems to be this almost random distribution of 
associations between Broad Street and Cardinals Terrace. I'm sure that 
the full NLPG might give some more clues. But we don't have access to that.

Mark