parkability

How easy is it to park, by Chicago ward / community area / ZIP — from open data.

A ward-wise-civic-tech direct metric source, sibling to chainshare. "Ease of parking" is a latent construct, so rather than one opaque score, parkability publishes clean component metrics that the Penlight explorer lets you weight yourself. It models parking as supply vs. scarcity.

Latest run: 13,748 off-street parking sites · 10,005 active permit-zone block faces · 50 wards / 77 community areas / 59 ZIPs.

Metrics

metric	meaning	geographies	toward "easier to park"
offstreet_parking_sites_per_sqkm	OSM off-street parking density (supply)	ward · community area · zip	higher
parking_311_complaints_per_sqkm	311 abandoned-vehicle + bike-lane parking complaints (stress)	ward · community area · zip	lower
parking_311_share_of_local_complaints_pct	parking complaints as a share of local 311 activity (reporting-controlled)	ward · community area · zip	lower
vehicles_per_household	ACS cars per household (demand)	ward · community area · zip	lower
permit_zone_block_faces_per_sqkm	residential permit-zone density (scarcity)	ward	lower

The share metric expresses parking complaints as a fraction of all local 311 activity (excluding the 311 INFORMATION ONLY CALL and Aircraft Noise Complaint bulk types, which otherwise swamp the denominator — aircraft-noise alone is ~1.2M reports, nearly all in the O'Hare ward). This controls for how much each area reports overall.

The permit-zone signal is deliberately central: the city designates these zones where residents compete for scarce street parking, so it's a clean "hard to park here" signal — and, unlike ticket counts, it is not confounded by enforcement intensity.

These are separate, weightable components, not one score, because they measure different things. Validated against real data:

• Permit-zone density peaks in dense North/Central wards (Lakeview, Lincoln Park, River North, Wicker Park) and bottoms out on the periphery — a clean scarcity signal.
• Off-street supply concentrates downtown (the Loop alone tags ~9,900 garage spaces).
• Car ownership is lowest in the transit-dense North lakefront / downtown (Ward 2: 0.56 veh/household) and highest in the SW/NW bungalow belt (Ward 23: 1.85) — the demand side.
• 311 parking complaints peak in NW/W working-class wards (Belmont Cragin, Portage Park) and are lowest in the dense, scarce-parking North/Central wards — even as a reporting-controlled share (9% of local 311 in the top wards vs ~2% in the Loop / Lincoln Park / River North). So this signal is genuinely anti-correlated with scarcity: it tracks vehicle-nuisance, a different dimension. (The truer "illegal parking" subtype, bike-lane complaints, is kept as a breakdown but is sparse — ~3.2k since 2023.) Methodology discloses this; weight accordingly.

Roadmap

• parking_tickets_per_1000_residents_2018 — illegal-parking density, clearly dated (the only comprehensive public set ends May 2018; refresh requires a FOIA to the Dept. of Finance) and enforcement-biased (normalized + caveated, never raw). Lower priority now that current 311 complaints cover the resident-report angle.
• Geocode permit-zone block faces → community-area / ZIP permit coverage (the source has no geometry, only built-in ward assignment, so those rollups need a one-time geocode).

Excluded by design: a single composite index; car-related crime (a weak, confounded proxy); metered-space supply (no clean official open feed).

Nomination

This metric was nominated by a resident of Ward 1 — see NOMINATION.md.

Outputs

data/processed/:

File	Grain
ward_parking_summary.json	one row per ward — the rollup Penlight ingests
community_area_parking_summary.json	one row per community area
zip_parking_summary.json	one row per ZIP
*.csv	tabular twins
metadata.json	provenance, per-metric geography coverage, caveats, audit

Run it

python -m parkability                 # live fetch (OSM Overpass + Chicago Data Portal + ACS)
python -m parkability --refresh       # force fresh fetch
python -m parkability \
  --parking-input    data/fixtures/sample_parking.json \
  --permit-input     data/fixtures/sample_permit_zones.json \
  --complaints-input data/fixtures/sample_311.json \
  --car-ownership-input data/fixtures/sample_car_ownership.json   # offline (tests use these)

The car-ownership metric needs a free Census API key — set CENSUS_API_KEY (https://api.census.gov/data/key_signup.html). Without it the run still produces the supply / scarcity / 311 metrics and just skips vehicles_per_household.

Standard library only — point-in-polygon, length, and area math are pure Python (geometry.py, measure.py); no shapely/geopandas. Tests: python -m pytest.

Bring your own polygons

parkability publishes fixed ward / community-area / ZIP rollups, but its underlying features are points, so it can aggregate to any polygons you hand it — no need to be one of the three bundled geographies. An aggregator (e.g. ward-wise / Penlight) can pass its own cells and get native per-polygon values instead of an areal estimate.

import json
from parkability.aggregation import aggregate_to_polygons

cells = json.load(open("my_polygons.geojson"))             # any FeatureCollection
values = aggregate_to_polygons(cells, id_field="cell_id")  # {cell_id: {metric: value, ...}}

or from the command line:

python -m parkability.aggregation --polygons my_polygons.geojson --id-field cell_id
python -m parkability.aggregation --publish-layers data/layers   # export the point layers + spec

What you get natively, point-in-polygon ÷ area: offstreet_parking_sites_per_sqkm, parking_311_complaints_per_sqkm, vehicles_per_household. Two metrics stay fixed-geography (a consumer should estimate them by area overlap instead): the 311 share needs the all-complaints denominator, which is aggregated server-side and not retained per point; and the permit-zone metric is ward-tagged address ranges with no point geometry. aggregation.AGGREGATION_SPEC declares this split, and --publish-layers writes the fine-grained GeoJSON point layers so a non-Python consumer can do the same point-in-polygon itself.

Sources & attribution

• Off-street parking: © OpenStreetMap contributors (ODbL), via Overpass.
• Permit zones: Chicago Data Portal — Residential Permit Zones u9xt-hiju (updated daily).
• Boundaries: ward + community area from ward-wise-civic-tech; ZIP from the Chicago Data Portal.

License

Code: MIT. Derived data inherits its sources' licenses (OSM data under ODbL).