add graph from travel network

[knaaptime]

import geodatasets
import geopandas as gpd
from geosnap import io
from libpysal.graph import Graph

df = gpd.read_file(geodatasets.get_path("geoda cincinnati"))

net = io.get_network_from_gdf(df)

Graph.build_travel_distance(df.set_index('ID'), net, 500, kernel=None).adjacency
>>> focal     neighbor
726907.0  726907.0      0.000000
          709415.0    385.609009
          709472.0    309.471985
          738259.0    346.858002
          738277.0    247.363007
                         ...    
703708.0  703537.0    349.321014
          703574.0    190.669006
          703629.0    370.070007
          703686.0    447.006989
          703708.0      0.000000
Name: weight, Length: 8075, dtype: float64


Graph.build_travel_distance(df.set_index('ID'), net, 500, kernel='triangular').adjacency
>>> focal     neighbor
726907.0  726907.0    1.000000
          709415.0    0.228782
          709472.0    0.381056
          738259.0    0.306284
          738277.0    0.505274
                        ...   
703708.0  703537.0    0.301358
          703574.0    0.618662
          703629.0    0.259860
          703686.0    0.105986
          703708.0    1.000000
Name: weight, Length: 8075, dtype: float64

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 85.4%. Comparing base (c97bf87) to head (f84bf89).
Report is 13 commits behind head on main.

Additional details and impacted files

@@           Coverage Diff           @@
##            main    #755     +/-   ##
=======================================
+ Coverage   85.3%   85.4%   +0.1%     
=======================================
  Files        149     150      +1     
  Lines      15914   15975     +61     
=======================================
+ Hits       13580   13641     +61     
  Misses      2334    2334             

Files	Coverage Δ
libpysal/graph/_network.py	100.0% <100.0%> (ø)
libpysal/graph/base.py	96.5% <100.0%> (+0.1%)	⬆️
libpysal/graph/tests/test_builders.py	100.0% <100.0%> (ø)

... and 1 file with indirect coverage changes

[knaaptime]

i think that windows failure is an issue in pandana's c lib, if I had to guess

[knaaptime]

probably better to call this build_travel_cost, since impedance isn't always distance

[martinfleis]

Cool!

[knaaptime]

this linter check is the bane of my existence

[jGaboardi]

the local pre commit gives me the willies. Ive gotten lost in rebasemergefastrorewardrewind too many times to like git making changes on my behalf

The way we set up pre-commit formatting is automatically changed, but not linting changes

[jGaboardi]

Your commit will fail and report linting errors if any linting violations are found.

[knaaptime]

yeah but sometimes i need to get the idea saved in git history before its looking pretty and I don't want my commit to fail (sorry formatting/linting the same in my head)

[jGaboardi]

in that case you can do git commit -m "..." --no-verify

[knaaptime]

but if just avoid setting it up entirely, I don't have to type those excruciatingly long 11 characters :P

[jGaboardi]

Having pandana as a hard dependency is a touch troublesome, no? For both Python 3.11 and Python 3.12 it took a considerably long time to get a compatible Conda Forge release, similar to numba. Maybe better to have as an optional dependency?

Also, it needs to be added to the pyproject.toml and environment.yml

[knaaptime]

it just consumes pandana--can only operate on a network if it's given, so no import/dependency (like xarray). so only changes are to the ci test files

[sjsrey]

Great addition!

[jGaboardi]

this one was your fault james!!! :P

LOL It was on purpose.