construct arc adjacency matrix directly from a geodataframe

[jGaboardi]

Following example given by @ljwolf from SciPy 2018 of constructing the arc adjacency matrix with a geopandas.GeoDataFrame.sindex for speed improvements.

Current procedure

1. extract vertices from networks lines (arcs).

• This means each euclidean segment within a network line, which means many small arcs for a sinuous line.

2. build adjacency list of nodes
3. create network arc representation based on vertices
4. build contiguity weights based on (3)

Proposed procedure

1. (when geopandas available), build contiguity weights based on each network arc (one record) in the geopandas.GeoDataFrame with the use of an sindex.
2. extract network vertices from the endpoints of (1)