Housing Types

Jan 2018 - Mar 2018

Algorithm to determine the housing type for all buildings in The Netherlands with Open Data

For this project we used only Open Data (Kadaster, BAG, and CBS). Housing types were determined by comparing the GIS buildig contour for all buildings within a neighbourhood. Kadaster GIS contour data (2D polygons) were imported and cleaned up with Python. The required 2D vector algebra to determine the distance between line segments and objects was implemented with custom Python classes. Combined with BAG building functions it was possible to classify each object into a standardized dutch housing classification of vrijstaand, tussenwoning, hoekwoning, twee onder een kap, and appartement (with subtypes). Results were verified using bagviewer.kadaster.nl.

My Contribution

Algorithm design
Data cleanup and processing pipeline
Implementation in Python
Exposure via API

Images and Screenshots

Example of contour data used to determine housing type:

External Links

Highlights

Classification of every building in The Netherlands
Housing type is not part of Open Data and was therefore a unique selling point for the client
Classification was integrated in the data ingestion pipeline for automatic updates

Users

multiple international energy corporations
local governments
several smaller clients

Technology Stack

Python
algorithms
AWS Lambda
AWS ApiGateway
PostGIS
Postgres