Historical house sale transactions can be downloaded from the Land Registry. However, the address data isn’t well mastered, and there are near misses on things like town and street names.
I spent a lot of time normalising and de-duplicating this data, in particular the addresses, so I could improve the quality of the results – and also recognise where a particular property had been sold multiple times. I have lots of SQL procedures as a result of this that I won’t bother publishing, but let me know if you’d like them.
For the rental data, I used python to connect to a rental property API. There was a pre-existing python API library, which I mangled a bit to get it to do what I wanted (this is data science not development!) – resulting in the following.
Finally, I wanted to visualise this data at the postcode and outcode level in Tableau. In order to do this I used the Information Lab’s UK Geocoding Pack. I extracted the postcode -> lat/long data into SQL to handle as part of my preprocessing, and also included the custom geocoding in Tableau so I could get the nice filled area maps in Tableau at the Outcode (e.g. N1, B23) level.