Is there a good, consistent normalization of city boundaries for meaningful comparisons? Like a 3-sigma population contour? Administrative regions seem really very distortionary.
Conversation
Replying to
Reasonable one I heard for defining a metropolis is how far people (regularly) commute. In air-travel era it becomes a bit tougher (eg, people commute between Cape Town <-> Johannesburg).
1
3
Replying to
I’m sure there’s a literature here but naively it seems like some kind of clustering algorithm is the way to go: merge nearest neighbors until every cluster is at least N people or M miles from the next one.
1


