Mapping new-build real estate prices
Predictive mapping, by statistical sector, of new-build price per m² in Brussels, for the urban charges framework.
Context
In the Brussels-Capital Region, urban charges (charges urbanistiques) are the financial contributions required from developers on real estate projects. Objectivising their geographic incidence requires knowing the value of new-build property, area by area.
Idea Consult needed a map, by statistical sector, of the new-build price per m² in Brussels. They had compiled by hand a database of prices from real estate listing sites, collected manually, as scraping does not comply with their terms of use.
What I did
- Geolocated each listing with the
phacochRpackage to assign it to its statistical sector. - Assembled a set of candidate explanatory variables from the regional open data portal datastore.brussels (income level, education level, etc.).
- Benchmarked several spatial interpolation algorithms under
mlr, compared by cross-validation, to predict the price per m² where data was missing. - Produced a choropleth map by statistical sector.
All in R: sf, tidyverse, mlr, ggplot2, leaflet.
Outcome
A complete, validated map of new-build price per m² by statistical sector, filling the gaps where data was missing: an objective basis to support the calculation of urban charges.