Mander is a little ditty that I've been doing to see how I can group parts of a road network into neighborhoods or districts. This is achieved by considering all intersections of a road network as subgraphs successively unioning these subgraphs based on a strategy that prioritizes the unioning of low-weight subgraphs (i.e. subnetworks which have a small total distance) and of subgraphs which are separated by short edges.
Creating and using districts like this can have a lot of different uses, such as helping to decide on delivery service areas, or informing solutions to the contentious issue of gerrymandering.
This project is not set up as a python module;
just a standalone script with dependencies specified in requirements.txt.
Postgres database connection in db.py.
Use run geo.py to execute the algorithm. You will be prompted for a name of an area, such as "Manhattan" or "Prince Edward Island" that is geocodable by OpenStreetMap.
The output will be a CSV that can be imported into a GIS as delimited text.
From there you can generate maps of the areas based on phases that the algorithm went through from 0 upwards.
There will be approximately 4 times as many subgraphs displayed in phase 0 than in phase 2.
Below are visualizations of districts created by the algorithm, mapped using QGIS. The dots represent intersections, while their colors represent district membership.
