postcodes

Compute postcode centroids

Install

Three options:

• You can use pip install -r requirements.txt into a virtual environment,
• If you have uv installed, run uv sync to get the the environment and dependencies installed at once
• If you have docker and the compose plugin, run docker compose build to create a local image with the dependencies ready

Running the script

The main.py script does the following:

• Downloads from CartoCiudad the geopackages for the three Valencian provinces into a data/provinces folder
• From each geopackage, extracts all the different street number points into CSVs per postcode into the folders data/postcodes/[alicante|castellon|valencia].
• Then for each postcode CSV it computes the centroid following these steps:
  • Compute a DBSCAN clustering using the parameters in dcac_postcodes/config.py
  • From the clusters that are not considered noise, compute the center and the closest point in the original dataset
  • Return the largest along with the number of points considered and their percentage
  • Store the result in data/postcodes.csv

The script is documented and running main.py -h will provide the settings available to change log level, force downloads, etc.

To use the docker recipe, just run docker compose run postcodes with the same options (so probably start with --help).