Related to my project with Data 4 Democracy, I’ve started compiling some scripts for spatial analysis. Particulary interesting (at least to me) is a bit of TSPLIB I adapted for use with the awesome osmnx library.
This is actually some synergy between the D4D project and the health violation prediction model I worked on. The idea here is to maximize “risk” coverage and minimize distance. Ideally, a reinforcement learning project. For now, though, to maintain my sanity, I’m having it as a single-agent problem.