UrbanAccessAnalyzer

Build street networks from OpenStreetMap data and compute multi-tier accessibility (isochrone-style access scores) to points of interest — schools, transit stops, green space, groceries, or any other OSM feature — per street edge, node, or aggregated to H3 hexagons.

The pipeline: an AreaOfInterest (geocoded or loaded from file) defines the study area; a StreetNetwork is cropped and built from a local or Geofabrik-downloaded .osm.pbf for a given profile (walk, bike, drive, all, ...), with connectivity/simplification handled internally (Polars + scipy.sparse.csgraph); PointsOfInterest are pulled from Overpass or supplied directly; and an AccessibilityAnalyzer computes per-edge/per-node access scores across configurable distance tiers, either whole-network or in memory-bounded H3-chunked passes for large areas.

Install

pip install -e .

# optional extras
pip install -e ".[plot]"       # matplotlib/folium/ipyleaflet mapping helpers
pip install -e ".[census]"     # WorldPop/country-level population data (pycensus)
pip install -e ".[geohierarchy]"  # aggregating street edges onto H3/other polygon layers
pip install -e ".[dev]"        # pre-commit, pytest, black, ruff

Requires Python >= 3.11. This project uses uv for dependency locking (uv.lock); uv sync works as an alternative to the pip install commands above.

Basic usage

from UrbanAccessAnalyzer import (
    AreaOfInterest,
    StreetNetwork,
    PointsOfInterest,
    AccessibilityAnalyzer,
)

aoi = AreaOfInterest.from_name("Cambridge, MA", buffer=500)

network = StreetNetwork.from_pbf(
    "massachusetts.osm.pbf",   # downloaded from Geofabrik if missing
    aoi=aoi,
    network_type="walk",
    simplify_distance=30.0,
)

points = PointsOfInterest.from_overpass("schools", aoi.gdf)

analyzer = AccessibilityAnalyzer(network, points)
node_access, edge_access = analyzer.run(distance_matrix=[400, 800, 1200])

access_gdf = analyzer.to_gdf(edge_access)   # scored street-edges GeoDataFrame
access_h3 = analyzer.to_h3(resolution=9)    # aggregated to H3 hexagons

For a one-call convenience wrapper around the same pipeline, see UrbanAccessAnalyzer.api.compute_accessibility. See examples/ for full notebooks (rural school access, walkability, green space, transit level of service).

Tests

pytest

Test fixtures use a small bundled .osm.pbf sample (tests/fixtures/); no network access or API keys are required to run the test suite.