Sketch of using networkx to calculate accessibility sheds

accessibility_shed.py

import networkx as nx
import numpy as np
import pandas as pd
import time
from dgna.utils import format_pandana_edges_nodes


# read in osm data, load into top level variable
osm_edges = pd.read_csv('./data/osm_edges.csv')
osm_nodes = pd.read_csv('./data/osm_nodes.csv')

# let's hardcode that we are looking jusr at transit accessibility
travel_speed_km = 4.8 # avg walk speed in kilometers per hour
osm_edges['weight'] = (osm_edges['distance'] / 1000) / travel_speed_km * 60

# prep nodes and edge for pdna
osm_edges, osm_nodes = format_pandana_edges_nodes(osm_edges, osm_nodes)

ids = osm_nodes.index.values
osm_nodes['id'] = ids

G = nx.DiGraph()
for i in ids:
	G.add_node(i, amenity=np.random.randint(20))

all_edges = osm_edges[['from_int', 'to_int', 'weight']].values
for edge in all_edges:
	G.add_edge(edge[0], edge[1], weight=edge[2])

print('preparing to calculate accessibility for {} geometries'.format(len(ids)))
start = time.time()

# get all amenities accessible within 15 minutes
result_rows = []
for i in ids:
	accessible = nx.single_source_dijkstra_path_length(G, i, cutoff=15)
	amenities_accessible = 0
	for a in accessible.keys():
		amenities_accessible += G.node[a]['amenity']

	result_rows.append((i, amenities_accessible))

# log a teaser of the outputs
print('example results:')
for r in result_rows[:5]:
	print('{}: {}'.format(*r))

end = time.time()
print("Completion time: %.2f s" % (end - start))