You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
An undocumented (and useful) feature of the Rook and Queen contiguity weights constructors is that passing in points results in the construction of voronoi polygons, which are then used to build contiguity weights. The current docs imply that only polygons can be passed. You need to read the source code to understand how points result in the identification of contiguous neighbors. I propose modifying the Rook and Queen docs to say something like.
To be fair, I don't think this behaviour should exist in the first place. Contiguity based on Voronoi polygons around points is not a Queen or Rook contiguity of point geometry. There's an explicit libpysal.weights.Voronoi constructor for that purpose. This is also the reason why we did not mirror this behaviour in Graph.build_contiguity.
i dont have a strong opinion about updating the docstring, but in general its a good idea to think of W as deprecated (even if it hasnt been, formally) and start using Graph whenever possible
also, we're not constructing the W from 'PySAL polygons', so the doc would need a bit more love anyway :P
An undocumented (and useful) feature of the Rook and Queen contiguity weights constructors is that passing in points results in the construction of voronoi polygons, which are then used to build contiguity weights. The current docs imply that only polygons can be passed. You need to read the source code to understand how points result in the identification of contiguous neighbors. I propose modifying the
Rook
andQueen
docs to say something like.The text was updated successfully, but these errors were encountered: