There are many different metrics that attempt to measure how weird-looking a district is. Whether measuring compactness, convexity, or bizarreness, all of the algorithms do a similar thing: They take in a set of geometries defining the districts of a plan, and return a number for each district indicating how much suspicion of gerrymandering should be given to the district.
A library of such metrics can be found in the python-mander repository in this project. This project provides a pipeline for applying such algorithms to real-world data and comparing to other existing algorithms.
- Download Data: data_downloader.py is a script for downloading a set of data from the internet to use as the test data
- Extract Data: The downloaded data is then processed to give it a consistent schema and projection.
- Compute Metrics: The values of a set of metrics are computed for comparison
- Display Results: Once the metrics have been computed, they must be understood. A set of pre-built visualizations have been provided to help the user identify patterns in the data. This is also helpful for explaining the different metrics to an unfamiliar audience. These are provided as a webpage. The officially maintained set of metrics will be hosted for now at http://www.gis.leg.mn/iMaps/metricViz/#.