subway-access 0.2.0

Spatial analysis toolkit for evaluating NYC subway accessibility, reliability, and equity gaps.

subway-access

subway-access is a Python toolkit for reproducible NYC subway accessibility analysis.

Authored by Blaise Albis-Burdige.

It is designed to measure neighborhood access to accessible stations with a small, transparent workflow that is easy to inspect, cache, analyze in memory, and extend.

What ships in the current line

The current package now includes a real public-data workflow:

• fetch MTA subway stations and ADA status from the public station catalog
• fetch public elevator and escalator availability history plus asset inventory
• fetch ACS tract-level demographics for a selected NYC study area
• cache a reusable local snapshot bundle
• analyze Euclidean first-pass accessibility gaps and rolling reliability
• export catchment GeoJSON, tract gap CSV, and station metrics
• run the snapshot and analysis flow from the installed CLI

The current scoring model is still intentionally a documented first pass: Euclidean catchments first, richer network comparisons next.

Why this exists

Official MTA and NYC data can tell you whether a station is nominally accessible, but the policy question is broader: which neighborhoods have weak access to accessible transit in practice?

This repo aims to grow into a reusable analysis toolkit rather than a notebook dump or trip planner.

Quickstart

Install:

pip install subway-access

Fetch a real official-data borough snapshot:

subway-access fetch-snapshot --geography borough --value Manhattan --cache-dir cache/manhattan

Then analyze the cached snapshot:

subway-access analyze-snapshot --cache-dir cache/manhattan --output-dir artifacts/manhattan

Examples

examples/ now follows the same self-contained project pattern used by nyc311. Each example folder has its own pyproject.toml, README.md, .gitignore, main.py, and tracked reports/ output.

Start with:

• examples/fetch-borough-snapshot/
• examples/borough-gap-analysis/
• examples/outage-reliability-report/
• examples/network-access-comparison/
• examples/example-template/

Python example

from pathlib import Path

from subway_access import analysis, models, pipeline

snapshot = pipeline.fetch_study_area_snapshot(
    models.AccessibilityQuery(geography="borough", value="Manhattan"),
    cache_dir=Path("cache/manhattan"),
)
catchments = analysis.generate_catchments(
    snapshot.stations,
    models.CatchmentRequest(minutes=10),
)
scores = analysis.score_accessibility(
    snapshot.stations,
    catchments,
    snapshot.demographics,
)
reliability = analysis.compute_reliability(
    snapshot.stations,
    snapshot.outages,
    models.TimeWindow(days=30),
)
gaps = analysis.analyze_gaps(scores)
print(len(gaps.records), len(reliability.records))

Current methodology

The current workflow is intentionally explicit and reproducible:

1. select a study area through nyc-geo-toolkit
2. fetch official MTA and Census sources into a local cache
3. load those cached files back into typed in-memory datasets
4. create Euclidean walk catchments using a fixed walking speed
5. test tract centroids against accessible-station catchments
6. compute tract need, rolling reliability, and station metrics
7. export publishable GeoJSON and CSV outputs

This is intentionally a first-pass Euclidean approximation, not a full network-isochrone model.

Documentation

• Hosted docs: subway-access.readthedocs.io
• Local preview: make docs
• Strict docs build: make docs-build

Quick links

Docs: Home, Getting Started, CLI Reference, Architecture, Python API, Contributing, Releasing, Changelog

Development

make install-dev
make test
make lint
make docs-build
make ci

License

MIT.