kartograf
This is a very basic web frontend for routing in New York City. It is not a production application and is intended solely for demonstrating predictive models involving transportation.
Getting started
First, install poetry if necessary. (On a Mac, use homebrew; on Linux, use your distribution's package manager or pipx.)
Then you can install the Python dependencies for kartograf:
poetry install
Next up, you can build a docker image for the routing service:
(cd nyc-routing ; docker build -t nyc-routing .)
Finally, you can run the routing service:
docker run --rm -t -p 5000:5000 nyc-routing
and the application:
poetry run python -m flask run --port 8000
Adapting the application
Use a local geolocation service
This application currently uses the public nominatim endpoint for geolocation, which is not suitable for heavy use. You can install your own nominatim service in a Docker container and then change the endpoint argument to resolve_address in app.py to refer to a locally-running service.
Add extra logic to map lookup
To integrate a predictive model into the app, add a call to the model itself (or a prediction API) from within the index function. You'll also need to pass any results that you want to render as keyword arguments to the render_template call in index and specify where you want to render them in map.html.