Blog

I've updated the site to include details of another level of administrative area, county electoral divisions. For example - Cambridgeshire postcodes

You can grab the data as CSV, KML, GeoJSON and Shape files. Let me know if you hit any problems.

In unrelated news, I've adjusted the blog subscription so you can subscribe just for blog posts about website updates or for all updates, since I occasionally rant about random subjects that most subscribers probably aren't interested in. If you've signed up already, you will only receive notifications about website updates. If you want to sign up for everything, currently you'll need to unsubscribe and sign up again. If you haven't signed up and would like to, head over to the sign-up page

I've uploaded the latest Land Registry website data to the website. Perhaps unsurprisingly, house prices are now falling and sales volumes don't look very healthy

I've added usage data for GB stations for the April 2022 to March 2023 time period. Whilst most stations still haven't returned to their pre-pandemic usage, the Elizabeth line appears to have been a huge success, with many of the stations along it achieving record numbers of passengers. The most used station is no longer London Waterloo, which has held the record for every year apart from 2021, but rather London Liverpool Street.

I've also updated the station stats page so you can explore the historic usage data

I've uploaded the latest ONS postcode data. There is an issue with ITL codes in the Northamptonshire area which I can't currently resolve but I will keep an eye out for updated data from the ONS. If you see any other problems, let me know

The server is currently churning through the latest house price data from the Land Registry. The trend continues, the number of sales is falling and prices are dropping with the annual percentage change likely to turn negative next month. The rise in interest rates finally seems to be having an effect.

I was complaining recently about the poor state of UPRN data. But it turns out I was mostly wrong and the problem was my poor research abilities (or things have changed for the better recently). A Disqus conversation led me to this page which provides the links between different identifiers.

These are all CSV files that can be imported into your favourite database. To get the street name of a UPRN, we need to download the UPRN -> USRN links file. Once we've got the USRN, we need to download the Road TOID -> USRN links file. Once we've got the TOID, we can link to the street name using the OS Open Names download (it's probably also possible to do this using the OS Open Roads download but that's a bit more difficult to process).

We're still not able to construct an open version of PAF but we are slowly creeping closer towards that. 

So I've pulled all that data into this site and started displaying it in a couple of places (like here and here) but I'm keen to hear suggestions for other potential uses for this data e.g. all the roads in a postcode area/district/sector, all the postcodes for a road. 

One thing to note. The OS provides a UPRN dataset and the ONS also provides a UPRN dataset. The OS version doesn't include postcodes but the ONS version does so is likely to be somewhat more useful.

Update - I'm not convinced the data provided in the linked identifiers is completely accurate. I noticed some strange results for some roads, so tried re-importing using the RoadLink TOID rather than the USRN (there's a useful diagram in this document that shows how everything links together). That gave better results but still not entirely accurate, this postcode shows two properties on the wrong street. I guess it's better than nothing but you're probably still better off purchasing PAF currently

Update 2 - Since I still wasn't happy with the results I was getting, I returned to the USRN to road TOID links and tried a different approach. The basic problem we have to deal with is that a single USRN can map to multiple road TOIDs. Most of these TOIDs aren't related to the USRN other than they intersect it. So I summed the distance from each UPRN to the centre point of each TOID. I'm now assuming the TOID that has the lowest value for this calculation is the TOID of interest. So far the results seem sensible. Let me know if you spot any oddities

I've uploaded the latest house price data for England and Wales to the website

Prices are heading down but the annual change remains in positive territory.

This seems like a big milestone. Electricity produced by renewables in the UK in 2023 is higher than electricity produced by fossil fuels. Hopefully the wind will continue to blow for the rest of the year and keep it that way. Source

I finally got round to importing schools data for England and Wales. I've not done a lot with it other than list nearby schools for individual postcodes. Let me know how you'd like to visualise or download the data.

One interesting thing I found in the schools data is the inclusion of the UPRN of the school. OK, it's interesting to me, possibly not to many other people. UPRNs uniquely identify addressable properties in Britain but although the data is freely available, it doesn't include the associated addresses. So it's kind of useless. I presume the reason we can't get the addresses is because the Royal Mail and its resellers make money from selling this data as PAF and if we could get a list of UPRNs and their addresses, PAF would be redundant. But this means no-one has much interest in UPRNs, because they don't serve any purpose for humans. 

We could try and build our own open source UPRN/address database, which is sort of happening at OSM. Run the query

node
  [ "ref:GB:uprn"]
  ["addr:street"]
  ({{bbox}});
out;

This currently doesn't return a huge amount of data and populating it is kind of difficult legally. I can find out my UPRN on Find My Address, but can I then use that to help populate OSM? Probably not, since the address data isn't open data. 

The government says we should be using UPRNs

"Systems, services and applications that store or publish data sets containing property and street information must use the UPRN and USRN identifiers."

But if everybody did this, all open property data would only include the UPRN and no address, which would make that data fairly unhelpful for humans. May I make the bold suggestion that it's time to make all this data open?

I've updated the Land Registry data on the website. The annual change is still in positive territory but prices look to be starting to head downwards. Taking into account inflation and wage increases, house prices are falling, but given the increases in mortgage costs, affordability is probably not