Generate a schematic map (“metro map”) for a given (transit) network graph using a Mixed Integer Programming approach. Part of the Generating Transit Maps project.

Work in progress, DO NOT USE THIS IN PRODUCTION! See todo.

npm version license chat on gitter

Installation & Requirements

Sadly, for now, you need a valid copy of the commercial Gurobi solver (free academic licenses) in order to run this project. In the near future however, this project will hopefully also support the open CBC solver. Stay tuned.

What you need:

Then, to install the package (for CLI usage), simply run:

npm install -g transit-map



You need a JSON graph representation of your transit network that looks like this example for the Berlin Metro (U-Bahn). You can then generate a transit map for the given graph by running:

cat graph.json | transit-map > output.svg

For further information on several CLI options/params, run:

transit-map --help

As a library

The module can be used as a JS library, documentation for this will follow.


Berlin (Metro)


Berlin metro network


Berlin metro map

Using this input graph. Running time ≈40sec.

Vienna (Metro)


Vienna metro network


Vienna metro map

Using this input graph. Running time ≈20sec.

Stockholm (Metro)


Stockholm metro network


Stockholm metro map

Using this input graph. Running time ≈20sec.

Lisbon (Metro)


Lisbon metro network


Lisbon metro map

Using this input graph. Running time ≈15sec.

Nantes (Tram)


Nantes metro network


Nantes metro map

Using this input graph. Running time ≈15sec.

Montpellier (Tram)


Montpellier tram network


Montpellier tram map

Using this input graph. Running time ≈20sec.

To do

Things I am still working on or someone should work on in the future, roughly in order of perceived importance:

  1. Properly document the current optimization algorithm.
  2. Add an example for full Berlin network (S-Bahn & U-Bahn), debug pre-processing with this (still fails with this network for some reason).
  3. Resolve todo comments in code.
  4. Write tests.
  5. Improve algorithm: “Smoother” edges, higher distance for parallel lines, implement station labels.
  6. Linearize objective function which would allow users to use non-commercial solvers like Cbc.
  7. Enable users to provide an old layout of the same network (or parts of it), the newly generated network should then have as few differences as possible to the older one.
  8. Further explore heuristical approaches to this problem.
  9. Write a paper. :D


If you found a bug or want to propose a feature, feel free to visit the issues page.

Built With

Share this project: