Looking at ONS UPRN assignment to postcodes in inner city Coventry (postcode sector CV2 4) and noticing a few errors anomalies:

1. a few UPRNs well outside the basic area of the sector (all appear to be historical)
2. misplaced UPRNs, such as this one for 44 Roman Road (UPRN 100070694394)

#UPRN #OfficeofNationalStatistics #PostCodes

Working with Australian and New Zealand address data? check out Matthew Proctor's database of AU & NZ Postcodes. https://matthewproctor.com/australian_postcodes #australia #NewZealand #PostCodes

Browsing rjw62's Mathmos QA site (https://osm.mathmos.net/addresses/pc-stats/PA/PA28/6/) I noticed an OSM postcode centroid for Machrihanish village was oddly displaced. Turns out this is because of a golf course tagged with an address.

This is a classic example of a mismatch between physical & postal addresses (the postal address is in the village as one might expect), and the actual clubhouse on the course has a different postcode.

#OpenStreetMap #addresses #postcodes #postalAddress #PhysicalAddress

@hmetcalfe

Version 2 with 7,264 polygons (#MSOA #Census small areas in #England and #Wales) and 1,531,286 points (active #postcodes in EW)

it's a loop over all MSOAs, considering only points and neighbors (using #rstats {spdep}) of one MSOA at a time, then binding

steadiliy below 3GB RAM, it's less than a second for most, with the occasional longer, in 10 mins it got beyond 1K polygons

with so many points it's possible there'll be some probs with intersections, to be dealt with afterwards