Making maps in code

Almost everything in the Built Environment happens somewhere. Buildings, streets, energy use, flood risk, heat, noise, who lives where: sooner or later your data has coordinates, and sooner or later someone asks you for a map. This section is the foundation for making that map in code, reproducibly, rather than by hand.

It grew out of the October 2026 special edition of the Coding Café, the #30DayMapChallenge kick-off, and it is written so that you can work through the challenge with it: one theme a day, 1 to 30 November. The day-by-day guide points every theme at the page that covers it.

How a map gets made¶

Whatever tool you use, a map in code goes through the same five steps. Most of the problems people hit are a missing or skipped step, usually the second one.

The pages¶

Anatomy of a map

Start here. One complete map, step by step, in Python and R. Everything else in the section is a deeper look at one of the five steps.

Vector data

Points, lines and polygons. What a GeoDataFrame is, and the handful of operations (filter, buffer, distance, spatial join, dissolve) behind most maps.

Projections

Why your map looks squashed. Coordinate reference systems, EPSG:4326 vs 28992 vs 3857, and the mistakes that silently give wrong distances and areas.

Finding data

OpenStreetMap, PDOK, CBS and Eindhoven open data. How to download from each, and what you must put on the map to credit it.

Map design

From a plot to a map. Visual hierarchy, colour, classification, legends, titles and credits, and a checklist to run before you post.

Grids, rasters and web maps

Beyond points on a plot. Hexagon grids, density surfaces, elevation rasters, basemaps and interactive maps.

Maps with AI

What an AI assistant gets wrong about maps, how to prompt so it does not, and how to check its work. Fits the 2026/27 Coding Café theme.

#30DayMapChallenge guide

All 30 themes, each with a technique, a page to read, and a data idea for Eindhoven or the Netherlands.

If you have one hour¶

Read Anatomy of a map and run its code on your own laptop.
Read the first half of Projections, up to The three you will meet.
Make the Day 1 Points map from the starter repo, then change one thing: the colour, the data, or the question it answers.
Before you share it, run the checklist at the end of Map design.

Setting up¶

You need either Python or R. You do not need both.

Python

No code

python -m venv .venv
source .venv/bin/activate          # Windows: .venv\Scripts\activate
pip install geopandas matplotlib mapclassify

Conda users: conda install -c conda-forge geopandas matplotlib mapclassify. geopandas brings shapely (geometry), pyproj (projections) and pyogrio (reading files) with it.

install.packages(c("sf", "ggplot2"))
# optional, used on some pages:
install.packages(c("tmap", "terra", "osmdata", "classInt"))

On Linux, sf needs the GDAL, GEOS and PROJ system libraries first (sudo apt install libgdal-dev libgeos-dev libproj-dev on Ubuntu). On Windows and macOS the CRAN binaries include them.