update 20-May-2022
wozwaardeloket.nl has changed its ways ...
Nothing here still works with the new setup

Previously I could get data of 10k objects within an hour, thus taking about 1000 hours for all NL data.
Running a few in parallel reduced the time a bit.

With the new setup I can get barely 600 objects per proces per hour.
I'll use that to finalise my 2021 dataset.
After that I'll publish the new scripts.

=============================

WOZ (=Waardebepaling Onroerende Zaak) values from wozwaardeloket.nl

Eventually you'll see here:

1.
Powershell script to extract WOZ data into .JSON files

2.
T-SQL script to load the data into database WOZ

3.
T-SQL script to load 'energyindex' data (CSV format from https://www.ep-online.nl/PublicData

4.
script to check which valid BAG ids are missing in WOZ
script to link Energieindex to WOZ (and BAG)
scripts to link WOZ to HUISV_AANBOD
scripts to link WOZ and some CBS data

eventually:
getting a model to predict the rent value of a property, especially social housing properties.
For social housing that's largely dependent on WOZ-value, sqm floor of rooms, and energyindex.

and
getting models to find anomalies in the available WOZ data


PLEASE NOTE:
Not all WOZ-data is available via the site wozwaardeloket.nl
The site has some sort of a fair use policy, but apparently no-one cares as early May 2021 (the time Dutch tax office and statistics office freeze their WOZ data), the site seemed overloaded, whereas end June I could extract data of about 50k properties an hour in 6 parallel running processes (still taking about 2 full weeks)

There are anomalies, e.g. about 150k  properties with not yet a 20200101 value while they have a 20190101 value (the difference between subsequent earlier years was only about 30k )

It stays a 'moving target' as people can and will object against WOZ-estimates by the local council, as there are local taxes based on that value, as well as national taxes on capital.
And the handling of objections carries over the national 'freeze' date.