Yinzone · Methodology & sources

How we got every number on the map

Sources, equations, weights, assumptions and limitations behind the parcel scores and housing-type fits. Weights are value judgments, not data. Zoning is a simplified interpretation, so verify it with the Zoning Administrator.

Decision support, not advice · Decision support only, not legal, financial or zoning advice. Zoning is a simplified reading of the City code; verify with the Zoning Administrator and a qualified professional before acting.

134

datasets

map layers

indicators

pillars

housing types

method refs

§01 · overview

How a parcel gets its numbers

Everything the map shows comes from public data run through deterministic code, with open weights you can change. Two AI components sit at the edges: one rates physical site fit, one explains. Neither decides legality or produces a score you can't trace. This page is generated from the same config and registry files the app reads, so it can't drift from what the UI does.

1 · Public dataobserved

134 datasets from City, County, WPRDC, Census, HUD, FEMA, EPA, USGS, PA agencies and OSM, pulled 2026-09-26/27.

2 · Build scriptscode

Clip to Allegheny County, clean known traps, write map overlays and per-parcel indicator shards. No model involved.

3 · Normalizecode

53 indicators → 0–100 by percentile rank or fixed linear thresholds. 100 = a good place to build.

4 · Weight + aggregatevalue

Open weights (a published JSON file) → sub-scores → 5 pillars → overall. Hazard gates cap Site Feasibility; deal-killers also multiply the overall.

5 · Legal + availabilitypolicy

Zoning use table (§911.02), current land use and deal-killer hazards multiply the overall score. Legality is read from code, never estimated.

6 · Site fit (Jev)model

A typed decision model rates physical fit per housing type from facts code already computed. Shown with its confidence.

7 · Verdict + pencilcode

Per housing type: red / yellow / green from pass/fail checks and a rough value-vs-cost screen. Never derived from the weighted score.

8 · Explain (chat)llm

Gemini explains on-screen facts with citations. Sentences with numbers not found in the facts are dropped.

Legendobserved Measured or recorded (counts, permits, sales, surveys)modeled Estimated by a model (EPA EJScreen air, HUD LAI, DOE LEAD, FEMA NRI)assumption Our assumption where data is missing (e.g. unit counts parsed from permit text)policy Current law or code (zoning use table)value A value judgment (weights, multipliers, directions)code Deterministic codemodel Typed decision modelllm Language model

Who decides what

Component	Kind	Decides	Never decides	If it fails
Pillar scorer (score.ts)	code	All five pillar scores, the overall score, hazard caps, zoning and availability multipliers	Anything not in pillars.config.json	n/a: pure function in the browser
Zoning gate (legal matrix + use table)	policy	Which approval pathway each housing type needs in each district	Whether a specific project will be approved	Unknown districts show "verify with the Zoning Administrator"
Jev site fit (15 housing types)	model	How well a housing type physically fits the lot, as a rubric level + probabilities + confidence	Legality, scores, any free text	Shown as unavailable; no fallback number is invented
Chat companion (Gemini)	llm	Wording of explanations; can call the rescore tool to show what a weight change does	Any score, legality, or number not present in the cited facts	Falls back to the screen's own notes

§02 · equations

The scoring equations

The exact math in apps/web/src/lib/pillars/score.ts (runs in the browser on every slider move) and scripts/pillars/build-indicators.ts (runs once at build time). Config version 2026-09-27.1.

E1 · Normalize: percentile (mid-rank)

p (v) s = 100 \cdot \frac{# { x < v } + \frac{1}{2} # { x = v }}{N} = {p (v) 100 - p (v) higher is better lower is better

$x$: every value in the reference set: all Allegheny County units in the source file (block groups, tracts, ZIPs), or all City parcels
$N$: size of the reference set
$s$: indicator score, rounded and clamped to 0–100

Ties share the middle of their rank, so a value shared by many units doesn't jump to the top or bottom.

E2 · Normalize: fixed linear scale

s = clamp_{[0, 100]} (100 \cdot \frac{v - v _{zero}}{v _{full} - v _{zero}})

$v_{zero}$: raw value that scores 0 (e.g. 2,400 m to a supermarket)
$v_{full}$: raw value that scores 100 (e.g. 400 m)

Used for distances, shares of lot area and other values with a meaningful absolute threshold. Category codes (lead line, legal pathway, parcel use) are used as-is.

E3 · Sub-score: weighted mean with renormalization

S_{k} = \frac{i \in A _{k} \sum w _{i} s _{i}}{i \in A _{k} \sum w _{i}} undefined if \frac{\sum _{i \in A_{k}} w _{i}}{\sum _{i \in I_{k}} w _{i}} < c_{m i n}

$w_{i}$: indicator weight from the config (0 = shown as context, not scored)
$I_{k}, A_{k}$: all indicators of sub-score k, and those with a value for this parcel; missing values are excluded, never filled
$c_{m i n}$: minimum coverage: 0.5 by default; Demand requires 0.75

E4 · Pillar: sub-scores, then hazard gates

P_{j} = min (\frac{\sum _{k} W _{k} S _{k}}{\sum _{k} W _{k}}, g \in G_{j} min cap_{g})

$W_{k}$: sub-score weight (scored sub-scores only); a pillar without sub-scores is one implicit sub-score
$G_{j}$: the pillar's gates whose conditions all hold for this parcel
$cap_{g}$: a ceiling such as 5 (half the lot in the floodway) or 50 (mostly 25%+ slope); cap 100 = warning only

Gates exist so an average can't hide a disqualifying hazard: great transit doesn't make a floodway buildable.

E5 · Overall: weighted geometric mean

O = exp (\frac{\sum _{j} ω _{j} ln max ( P _{j} , 1 )}{\sum _{j} ω _{j}}) P_{j} \leftarrow impute_{j} if missing

$ω_{j}$: pillar weight (0–3 in the navbar Weights popover; published default 1)
$max (P_{j}, 1)$: floor of 1 keeps a zero pillar from zeroing the whole product
$impute_{j}$: City 25th-percentile score for that pillar, used when it lacks data (and flagged): Demand 34 · Site Feasibility 70 · Affordability 42 · Access to Opportunity 58 · Climate & Environment 56

Weighted geometric mean limits how much a strong pillar can offset a weak one (UN HDI, 2010 switch; OECD/JRC Handbook 2008). Scores are floored at 1 so a zero doesn't zero the product. Arithmetic is available as a toggle.

E6 · Final parcel score

Score = O \times m_{legal} \times m_{avail} \times m_{hazard}

$m_{legal}$: multiplier for the easiest legal pathway among detached, townhouse, two-, three- and multi-unit homes, or for one type the user picks in the panel (§4)
$m_{avail}$: multiplier for what's on the lot now: vacant 1.0 … park, rail, right-of-way 0.05
$m_{hazard}$: lowest overall factor among the fired deal-killer gates (floodway, sliver lot, mines), else 1 (§4)

Multiplied, not averaged: a parcel where housing isn't permitted, or sits in a floodway, can't be rescued by good access.

E7 · Contribution shown in the panel

share_{i} = \frac{W _{k}}{\sum _{k^{'}} W _{k^{'}}} \cdot \frac{w _{i} s _{i}}{\sum _{i^{'} \in A_{k}} w _{i^{'}}}

The shares of one pillar add up to the pillar score before any gate cap, so the parcel panel can say exactly which indicator added how many points.

E8 · Weight sensitivity (rank stability)

ω^{'} \sim Dirichlet (α \frac{ω}{\sum _{j} ω _{j}}), α = 47, 200 draws

$P_{10}, P_{50}, P_{90}$: percentiles of the overall score O(ω′) across the draws, shown as the range in the parcel panel

Dirichlet draws around the current pillar weights. Concentration 47 gives a 0.2 weight a standard deviation of about 0.058, matching the ±0.1 uniform weight ranges used in the JRC audit of the Global Innovation Index.

Worked example

computed live by score.ts

Demand 72Site Feasibility 85Affordability 30Access to Opportunity 64Climate & Environment 58

O_{geo} O_{arith} Score = exp (\frac{ln 72 + ln 85 + ln 30 + ln 64 + ln 58}{5}) = 58.4 = \frac{72 + 85 + 30 + 64 + 58}{5} = 61.8 = 58.4 \times special exception 0.94 \times occupied building 0.9 = 49.4

The weak Affordability pillar (30) pulls the geometric mean 3.4 points below the arithmetic one. That is the intended effect: strengths elsewhere only partly offset a real weakness.

§03 · pillars

The five pillars, indicator by indicator

Read straight from pillars.config.json. Every indicator, sub-score and pillar is 0–100, where 100 means this is a good place to build new housing and 0 means a poor one. Area indicators are percentile-ranked across all Allegheny County units in their source file (block groups, tracts or ZIPs); parcel indicators use fixed thresholds. Weights are value judgments, not data.

Demand

weight 1 · 5 scored / 8 indicators · impute 34

Is there demand for new homes here? Household and job growth, low vacancy, an active sales market, rising rents and recent construction score high.

value Value judgment: Demand measures market strength (sales activity, low vacancy, new construction), so it correlates strongly (0.77) with the Reinvestment Fund market type that is otherwise kept at weight 0. At full weight it pulls the overall toward strong markets; Housing Need pulls toward high-need places. Lower Demand's weight to favor need.

Indicator	Geo	Weight	Evidence	Normalization	Input	Why
Household change, 2020 Census to ACS 2020–24 demand_household_growth	tract	1	observed	percentile ↑ vs. county units	table: allegheny_tract_household_growth_2020_2024.csv → hh_change_pct	Household growth is the core of HUD market-analysis demand. Weak signal: only 35 of 394 tracts change by more than the ACS margin of error, so it carries weight 1. Changes inside the ACS margin of error are treated as zero change.
Home sales per 100 owner-occupied units, 2024–25 demand_sales_turnover	tract	1.5	observed	percentile ↑ vs. county units	table: derived_tract_turnover.csv → sales_per_100_owner_units	Arm's-length sales relative to the owner-occupied stock they come from. Dividing by all housing units measured homeownership share instead of market activity (round-1 review). Blank where a tract has fewer than 150 owner-occupied units.
Residential vacancy (USPS) demand_low_vacancy	tract	2	observed	percentile ↓ vs. county units	area: usps-vacancy.geojson → res_vacancy_pct	Low vacancy means existing homes are being absorbed. Blank for tracts with fewer than 200 residential addresses.
Job change, 2019 to 2023 demand_job_growth	block group	0.5	observed	percentile ↑ vs. county units	table: allegheny_jobs_demand_bg_2023.csv → jobs_change_pct	Employment growth drives household formation (LEHD LODES). Blank where a block group had fewer than 250 jobs in 2019, since small bases give meaningless growth rates.
Rent growth, last 12 months (Zillow ZORI) demand_rent_growth	ZIP	0 · context	observed	percentile ↑ vs. county units	area: market-zip.geojson → zori_yoy_pct	Context only (weight 0): 12-month ZIP rent change is lowest exactly where new supply lands, and it ran opposite to the rest of Demand (round-2 review).
New homes permitted within 800 m, 2019–26 (log units) demand_new_construction	parcel	1.5	assumption	percentile ↑ vs. City parcels	sum_within within 800 m: pgh_new_residential_permits_classified.csv → units_guess (log1p)	Revealed demand: builders are building nearby. Counts homes, not permits, on a log scale so one 224-unit building counts more than one house but doesn't swamp everything, and includes apartment buildings, which the City files as 'Commercial'. Unit counts are parsed from permit text (assumption where missing: 1). Several permits on one parcel (phases, trades) count once, at the largest unit count.
Market Value Analysis type (A strongest … J most distressed) demand_market_strength	block group	0 · context	value	linear 1 → 0, 10 → 100	area: market-mva.geojson → mva	Default weight 0: rewarding strong markets steers housing away from distressed neighborhoods. Available for users who choose to weigh it.
Median home sale price, 2024–25 (census tract) demand_median_sale_price	tract	0 · context	observed	percentile ↑ vs. county units	table: allegheny_tract_sales_2019_2025.csv → median_2024_2025	Context only (weight 0): what homes nearby sell for, close to what an appraiser's comparable sales would show. It feeds the pencil check, not the score. Valid residential sales only; blank under 10 sales.

Site Feasibility

weight 1 · 5 scored / 11 indicators · impute 70

Can this lot physically take new housing? Flooding, steep slopes, landslides, undermining and known lead service lines score low. Zoning legality is applied per housing type, not averaged in.

Indicator	Geo	Weight	Evidence	Normalization	Input	Why
Share of lot in the FEMA regulatory floodway site_floodway_share	parcel	0 · context	observed	linear 1 → 0, 0 → 100	overlap: flood-zones.geojson	Used only as a gate: half or more of the lot in the floodway caps Site Feasibility at 5 and multiplies the overall score by 0.2; any part caps it at 40 and multiplies the overall by 0.6. Never averaged.
Share of lot in the 100-year floodplain site_sfha_share	parcel	3	observed	linear 1 → 0, 0 → 100	overlap: flood-zones.geojson	FEMA Special Flood Hazard Area; triggers the City floodplain overlay (Ch. 906).
Share of lot at 25%+ slope (City map, 2018) site_steep_slope_share	parcel	3	observed	linear 1 → 0, 0 → 100	overlap: slope25.geojson	Triggers the City steep-slope overlay (Ch. 906) and raises foundation cost. Half or more of the lot at 25%+ slope caps Site Feasibility at 50 (was three-quarters until 2026-09-27; any development in SS-O needs a Planning Commission hearing, §906.08.C.1). Not an overall multiplier: a hackathon SME described slope as added cost, not a deal-killer.
Share of lot in a City landslide-prone area site_landslide_prone_share	parcel	2	observed	linear 1 → 0, 0 → 100	overlap: landslide-prone.geojson	Triggers the City landslide-prone overlay (Ch. 906).
Share of lot on mapped landslide deposits (county: recent, old or red-bed) site_recent_landslide_share	parcel	1	observed	linear 1 → 0, 0 → 100	overlap: landslide-susceptibility.geojson	County susceptibility mapping. 'Recent' alone covered 0.2% of parcels and none of the City's incident points; adding old and red-bed deposits covers half of them (round-1 review).
Share of lot over mapped mines (City undermined area) site_undermined_share	parcel	1	observed	linear 1 → 0, 0 → 100	overlap: undermined.geojson	Triggers the City undermined overlay (Ch. 906); subsidence risk. Half or more of the lot over mapped mines caps Site Feasibility at 70 and multiplies the overall score by 0.9: anything larger than a single house needs a mine investigation first (§906.05).
Known lead or galvanized service line at this address site_lead_line	address	0 · context	observed	code (already 0–100)	nearest_attr ≤ 30 m: lead-service-lines.geojson → status	Flag only (weight 0): the data has only lead/galvanized/unknown records, so it could only subtract, and it describes the existing house's line, not new construction.
FEMA NRI inland flooding score site_inland_flood_rating	tract	0 · context	modeled	percentile ↓ vs. county units	area: weather-risk.geojson → inland_flood_score	Context only (weight 0). FEMA's National Risk Index score multiplies expected annual loss by a community risk factor built from social vulnerability, so it partly measures who lives there, not just the hazard (round-2 equity review). Parcel-level floodplain and surface-heat indicators carry these hazards instead.
Easiest legal pathway for mainstream housing types (current code) site_legal_pathway	parcel (zoning district)	0 · context	policy	code (already 0–100)	legal_pathway: typology-district-matrix.json	Not averaged into Site Feasibility. It multiplies the overall score instead (see the legal block in this file).
Site availability (what is on the parcel now) site_parcel_use	parcel	0 · context	observed	code (already 0–100)	parcel_use:	Not averaged; it multiplies the overall score (availability block). From the County assessor land use and class, owner category, the City vacancy flag, zoning (P, EMI) and overlap with mapped parks. First matching rule wins.
Lot area (for the sliver-lot check) site_lot_area	parcel	0 · context	observed	linear 0 → 0, 600 → 100	parcel_attr: → Shape__Area	Not averaged. Lots under 600 sq ft (slivers) cap Site Feasibility at 30 and multiply the overall score by 0.3, unless they are condo units.

Gate: fires when	Cap	Overall ×	Flag shown
Share of lot in the FEMA regulatory floodway score < 51	≤ 5	0.20	Half or more of the lot is in the FEMA regulatory floodway
Share of lot in the FEMA regulatory floodway score < 100	≤ 40	0.60	Part of the lot is in the FEMA regulatory floodway
Share of lot in the 100-year floodplain score < 51	≤ 50	—	Half or more of the lot is in the 100-year floodplain (City floodplain overlay applies)
Share of lot at 25%+ slope (City map, 2018) score < 51	≤ 50	—	Half or more of the lot is 25%+ slope (steep-slope overlay: Planning Commission hearing, §906.08)
Share of lot in a City landslide-prone area score < 51	≤ 54	—	Half or more of the lot is landslide-prone (geotechnical study required)
Share of lot at 25%+ slope (City map, 2018) score < 51 AND Share of lot in a City landslide-prone area score < 51	≤ 45	—	Steep and landslide-prone
Share of lot over mapped mines (City undermined area) score < 51	≤ 70	0.90	Over mapped mines: anything larger than a single house is prohibited until a mine investigation clears it (§906.05)
Known lead or galvanized service line at this address score < 50	flag only	—	Known lead or galvanized service line on this address
Lot area (for the sliver-lot check) score < 100 AND Site availability (what is on the parcel now) ≠ 3	≤ 30	0.30	Sliver lot under 600 sq ft

Affordability

weight 1 · 7 scored / 11 indicators · impute 42

How much unmet need for homes people can afford is here, measured by households: low-income renters who are cost-burdened and the share of renters with extremely low incomes. Displacement pressure is shown alongside as a flag, not scored.

value Value judgment: greater need raises the score. Displacement pressure is shown, not scored: research finds new supply tends to relieve nearby rents and displacement (Asquith, Mast & Reed 2023; Pennington 2021), and the right response to pressure is usually income-restricted homes and tenant protections, not avoiding the area. Round-1 review found that scoring price growth pushed down the poorest, mostly Black neighborhoods.

affordability Housing need · weight 1displacement Displacement pressure (shown, not scored) · weight 0 (shown as a flag, not scored)

Indicator	Geo	Weight	Evidence	Normalization	Input	Why
Low-income renters paying 30%+ of income on housing (CHAS) afford_lowinc_renter_burden · affordability	tract	2	observed	percentile ↑ vs. county units	area: chas-cost-burden.geojson → renter_lowinc_costburdened_pct	HUD's 30% cost-burden standard for households at or below 80% AMI. More burdened renters = more need for homes they can afford. Left out in tracts where more than 30% of residents are enrolled in college, because student households read as low-income.
Renters at or below 30% of area median income (CHAS) afford_eli_renter_share · affordability	tract	3	observed	percentile ↑ vs. county units	area: chas-cost-burden.geojson	Extremely low-income renters are the households the private market serves least. Tract share of renter households; blank under 50 renter households. Student-heavy tracts can inflate it (not yet corrected). Left out in tracts where more than 30% of residents are enrolled in college, because student households read as low-income.
Median gross rent vs. rent affordable at 50% AMI (3-person household) afford_rent_vs_ami · affordability	block group	0 · context	assumption	linear 0.8 → 0, 2 → 100	area: housing-costs.geojson → median_gross_rent	Context only (weight 0): rent relative to AMI measures how expensive the market is, not how many households are struggling (round-1 review). 1242.5 = 30% × $49,700 (HUD FY2026 Pittsburgh 50% AMI, 3 persons) ÷ 12.
Renters paying 30%+ of income on rent (ACS) afford_rent_burden · affordability	block group	1	observed	percentile ↑ vs. county units	area: housing-costs.geojson → rent_burden_pct	All renters, not just low-income; ACS margins of error are large at block group level. Left out in tracts where more than 30% of residents are enrolled in college, because student households read as low-income.
Housing + transportation cost share, median-income family (HUD LAI v3) afford_housing_transport · affordability	tract	0 · context	modeled	linear 45 → 0, 75 → 100	area: location-affordability.geojson → ht_median_family	Context only (weight 0): modeled for a fixed household, barely varies within the City, and duplicates the transit and jobs access already in Access. CNT's H+T benchmark treats ≤45% of income as affordable.
Rent growth, last 5 years (Zillow ZORI) afford_rent_growth_5yr · displacement	ZIP	1	observed	percentile ↓ vs. county units	area: market-zip.geojson → zori_5yr_pct	Fast rent growth is a displacement pressure (Urban Displacement Project typology uses rent change).
Displacement Risk Ratio, 2019/20 (Reinvestment Fund) afford_displacement_ratio · displacement	block group	2	observed	percentile ↓ vs. county units	area: market-mva.geojson → drr_1920	Sale-price growth relative to residents' incomes (Reinvestment Fund). Data end 2019/20, so low weight; tract price growth carries the current signal.
Median home sale price change, 2020–21 to 2024–25 afford_price_growth · displacement	tract	1	observed	percentile ↓ vs. county units	table: allegheny_tract_sales_2019_2025.csv → chg_pct_pooled_reliable	Median sale price change from pooled 2020–21 to pooled 2024–25 (county sales file). Blank unless both pools have at least 10 arm's-length sales.
Eviction filings per 100 renter households, 2025 afford_evictions · displacement	ZIP	0 · context	observed	percentile ↓ vs. county units	table: pittsburgh_evictions_zip_2025.csv → filings_per_100_renter_hh	Context only (weight 0): Eviction Lab publishes only ZIP-level filings for Allegheny County, and the ZIP value barely relates to the other displacement signals (r = −0.05).
Home energy burden, households at or below 80% AMI (DOE LEAD 2022) afford_energy_burden · affordability	tract	1	modeled	percentile ↑ vs. county units	table: allegheny_energy_burden_tract_2022.csv → energy_burden_lowinc	Energy bills are part of housing cost. Modeled from ACS 2022 PUMS, not utility bills. Left out in tracts where more than 30% of residents are enrolled in college, because student households read as low-income.
Residents enrolled in college (ACS 2020–24) afford_student_share · affordability	tract	0 · context	observed	linear 0 → 0, 1 → 100	table: allegheny_tract_college_enrollment.csv → pct_enrolled_college	Context and flag only. Over 30% triggers the student adjustment: the income-based need indicators are left out for the tract (a methodological choice). Includes dorm residents.

Gate: fires when	Cap	Overall ×	Flag shown
Displacement Risk Ratio, 2019/20 (Reinvestment Fund) score < 10	flag only	—	Sale prices outpaced local incomes by 2019/20 (Reinvestment Fund DRR, top 10% of the county): income-restricted homes and tenant protections matter most here
Median home sale price change, 2020–21 to 2024–25 score < 10	flag only	—	Strong recent home-price growth (2020–21 to 2024–25): income-restricted homes and tenant protections matter most here
Residents enrolled in college (ACS 2020–24) score ≥ 31	flag only	—	Over 30% of residents are college students: income-based need measures are left out here because student households read as low-income

Access to Opportunity

weight 1 · 15 scored / 15 indicators · impute 58

Would new residents reach daily needs? Walking distance to groceries, health care, schools and services, transit service, and jobs reachable by transit.

Indicator	Geo	Weight	Evidence	Normalization	Input	Why
Weekday transit trips within 800 m, distance-weighted access_transit	parcel	3	observed	percentile ↑ vs. City parcels	decay_sum: transit-stops.geojson → trips_wd	Transit Score pattern: trips × mode weight × distance decay.
Walking distance to nearest supermarket access_grocery	parcel	2	observed	linear 2400 → 0, 400 → 100	nearest: places-groceries.geojson	USDA SNAP-authorized supermarkets and super stores only. The broader 'full_grocery' tier included convenience/dollar stores and stale permit rows (round-1 review). Walk Score decay: full credit within 400 m, none beyond 2.4 km. Straight-line distance.
Walking distance to nearest small grocery store access_small_grocer	parcel	0.5	observed	linear 1600 → 0, 400 → 100	nearest: places-groceries.geojson	SNAP 'Grocery Store' type: smaller grocers that carry staples.
Shops and restaurants within 800 m access_shops_restaurants	parcel	2	observed	percentile ↑ vs. City parcels	count_within within 800 m: commerce-density.geojson	A main-street proxy (WPRDC food-facility and business points). Raised agreement with Walk Score from 0.85 to 0.90 in the round-1 review.
Walking distance to nearest pharmacy access_pharmacy	parcel	1	observed	linear 2400 → 0, 400 → 100	nearest: places-pharmacies.geojson	Retail pharmacies from the CMS NPPES registry (2026), with closed Rite Aid stores removed.
Walking distance to nearest school access_school	parcel	1	observed	linear 2400 → 0, 400 → 100	nearest: places-schools.geojson	Proximity only; school quality is deliberately not scored.
Walking distance to nearest park access_park	parcel	1	observed	linear 1200 → 0, 200 → 100	nearest: parks.geojson	Distance to the nearest park boundary. Tightened (full credit within 200 m, none beyond 1.2 km) because nearly every parcel was near some park.
Distance to nearest health center, clinic or hospital access_health	parcel	1	observed	linear 3200 → 0, 800 → 100	nearest: places-health-centers.geojson + places-clinics.geojson + places-hospitals.geojson	HRSA health-center sites, NPPES urgent-care/primary-care/community clinics, and hospitals.
Walking distance to nearest child care access_child_care	parcel	1	observed	linear 2400 → 0, 400 → 100	nearest: places-child-care.geojson
Walking distance to nearest library access_library	parcel	0.5	observed	linear 2400 → 0, 400 → 100	nearest: places-libraries.geojson
Walking distance to nearest food pantry or walk-up food distribution access_food_assistance	parcel	0.5	observed	linear 2400 → 0, 400 → 100	nearest: places-food-banks.geojson	Greater Pittsburgh Community Food Bank network (Apr 2025). Drive-up distributions are excluded because they are scheduled events, not places you can walk to.
Distance to nearest dentist access_dentist	parcel	0.5	observed	linear 2400 → 0, 400 → 100	nearest: places-dentists.geojson	OSM + County Assets. No data on which dentists accept Medicaid.
Walking distance to nearest community center access_community_center	parcel	0.5	observed	linear 2400 → 0, 400 → 100	nearest: places-community-centers.geojson
Distance to nearest senior center access_senior_center	parcel	0.5	observed	linear 2400 → 0, 400 → 100	nearest: places-senior-centers.geojson	Matters most for older adults; the older-adult preset raises Access overall.
Jobs reachable within 30 minutes by transit access_jobs_transit	block group	2	modeled	percentile ↑ vs. county units	area: jobs.geojson → transit_jobs_30	Cumulative-opportunity accessibility (EPA Smart Location Database style).

Climate & Environment

weight 1 · 7 scored / 8 indicators · impute 56

Would homes here be low-carbon and healthy? Two sub-scores that often pull in opposite directions: carbon (dense, walkable places mean less driving) and local environment (air, heat, tree canopy). Kept separate so the trade-off stays visible.

carbon Carbon (location efficiency) · weight 1local_env Local environment · weight 1

Indicator	Geo	Weight	Evidence	Normalization	Input	Why
Fine particulate matter (PM2.5), PA percentile climate_pm25 · local_env	block group	2	modeled	percentile ↓ vs. county units	area: air-quality.geojson → pm25_pct	EJScreen v2.3 modeled annual exposure, re-ranked within the county.
Traffic combustion (mean of diesel PM, NO2 and traffic proximity percentiles) climate_traffic_combustion · local_env	block group	1	modeled	percentile ↓ vs. county units	area: air-quality.geojson → dslpm_pct,no2_pct,ptraf_pct	Three correlated traffic indicators averaged first so traffic counts once.
Industrial air toxics (RSEI), PA percentile climate_air_toxics · local_env	block group	1	modeled	percentile ↓ vs. county units	area: air-quality.geojson → rsei_air_pct
Industrial PM2.5 emissions nearby, distance-weighted (ACHD, ≤2021) climate_industrial_emissions · local_env	parcel	1	observed	percentile ↓ vs. City parcels	decay_sum: achd_facilities_latest.csv → pm25_total_tpy	PM2.5 tons/yr (filterable + condensable) from current permitted facilities, full weight within 1 km, fading to zero at 10 km so Clairton Coke Works reaches the Mon Valley neighborhoods. Excludes NRG Cheswick, which stopped generating on March 31, 2022. Emissions, not measured exposure; the inventory ends in 2021.
Modeled annual household VMT, median-income family (HUD LAI v3) climate_household_vmt · carbon	tract	2	modeled	percentile ↓ vs. county units	area: location-affordability.geojson → vmt_median_family	Transportation-carbon proxy. HUD models VMT for a fixed household (median income, 4 people, 2 commuters), so it varies only with location efficiency. Inputs 2012–16, 2010 tracts.
FEMA NRI heat wave score climate_heat · local_env	tract	0 · context	modeled	percentile ↓ vs. county units	area: weather-risk.geojson → heat_wave_score	Context only (weight 0). FEMA's National Risk Index score multiplies expected annual loss by a community risk factor built from social vulnerability, so it partly measures who lives there, not just the hazard (round-2 equity review). Parcel-level floodplain and surface-heat indicators carry these hazards instead.
Summer surface heat, parcel (Landsat 8, 4-scene anomaly vs county median) climate_surface_heat · local_env	parcel (30 m pixel at centroid)	2	observed	percentile ↓ vs. City parcels	table: pgh_parcel_lst_canopy.csv → lst_anom_c_4scenes	Mean of (LST − scene county median) over 4 clear summer scenes, 2023–2025. Surface, not air, temperature; ranked, not thresholded.
Tree canopy around the parcel (block-group mean, NLCD 2021) climate_tree_canopy · local_env	block group	1	observed	linear 0 → 0, 50 → 100	table: allegheny_bg_canopy_impervious_lst.csv → canopy_pct_mean	Block-group mean canopy. The single 30 m pixel at the parcel centroid read 0 for a quarter of parcels (round-1 review).

Weight presets

navbar → Weights

Preset	Demand	Site Feasibility	Affordability	Access to Opportunity	Climate & Environment	Note
equal	1	1	1	1	1
family	1	1.5	1	2	1
older_adult	1	2	1	2.5	2
climate_first	1	1	1	1.5	3
affordability_first	1	1	3	1.5	1	Ranks unmet housing need highest; favors neighborhoods where low-income renters are most burdened.
market_first	3	1.5	0.5	1	0.5	Ranks strong markets highest. Correlates negatively with minority share (round-2 equity review), so it steers new housing away from disinvested neighborhoods.

§04 · multipliers

Zoning, site-availability and hazard multipliers

The three factors in E6. All are value judgments about how much approval risk, existing use and deal-killer hazards should cost, applied on top of policy and observed facts.

Legal pathway (m_legal)

indicator site_legal_pathway · border 30 m

Whether current zoning lets you build the brief's mainstream housing types (detached, townhouse, two-unit, three-unit, multi-unit) on this parcel. The overall score is multiplied by the factor for the easiest legal pathway among them. Legality is read from the City use table (§911.02), not estimated.

Code	Level	×	Note
0	Housing allowed by right	1.00
1	Housing allowed with Zoning Administrator approval	0.98
2	Housing needs a ZBA special exception	0.94	ZBA approved 90% of special exceptions in 2023–26 (n=30), counting only posted decisions.
3	Housing needs conditional use (Planning Commission + Council)	0.92
4	Housing not permitted, but the parcel touches a district that allows attached or multi-unit homes	0.35	Exceptions happen at district edges: a use variance or a map amendment extending the neighboring district. Since 2015, Council adopted 42 of 49 site rezonings and defeated none (7 held, withdrawn, pending or expired). But few housing use variances are in districts that don't allow housing.
5	Housing not permitted: needs a rezoning or use variance	0.20	Not legally impossible, but the slowest, least certain path.
6	Planned district: depends on the site's approved plan	0.96
7	Outside City zoning jurisdiction (Mount Oliver Borough)	1.00
8	Zoning district not in our use table: verify with the Zoning Administrator	0.80
9	Hillside (H) district: detached house only, with ZA approval and Site Plan Review	0.85

Site availability (m_availability)

indicator site_parcel_use

What is on the parcel now. Vacant land and surface parking are the easiest places to add housing; occupied buildings, institutions and large buildings are progressively less likely to be redeveloped; parks, cemeteries, rail and rights-of-way are not development sites. The overall score is multiplied by the factor below.

Code	What's on the lot	×	Note
0	Vacant land or surface parking	1.00
5	Existing occupied building	0.90	Building here means replacing or adding to what's there. Where displacement pressure is flagged, replacing occupied homes risks direct displacement.
6	Large existing building (20+ apartments, office tower, hotel, shopping center)	0.70	Unlikely to be redeveloped for new housing soon.
7	Institution (university, school, hospital or care home, church, occupied government building)	0.60	Institutional land is rarely released for housing; when it is, it goes through master plans (EMI districts) or surplus sales.
3	Unit in an existing condominium building	0.50	Scores describe the building's location. Building new housing here would mean redeveloping the whole building lot.
1	Park, cemetery or public open land: not an available site	0.05	Converting parkland needs court approval (PA Donated or Dedicated Property Act); cemeteries are effectively permanent.
2	Active railroad: not an available site	0.05
4	Right-of-way or air rights: not an available site	0.05

Deal-killer hazards (m_hazard)

Site Feasibility gates with an overall factor

A few Site gates are deal-killers, not just a low Site score: most of the lot in the floodway, a sliver lot, or mapped mines under anything larger than a house. Capping only the Site pillar let the other four pillars average them away (a fully undermined lot moved the overall about 2%). Those gates carry an `overall` factor that multiplies the overall score, like zoning and availability; the lowest fired factor applies. Hackathon SMEs (2026-09-27) called undermining and environmental conditions up-front deal killers. Undermining gets a mild factor (×0.9) because about 30% of City parcels (43,001 of 142,365) are over mapped mines and the layer gives no depth of cover: it means a mine investigation, not a ban. The factors are value judgments.

Gate: fires when	Overall ×	Flag shown
Share of lot in the FEMA regulatory floodway score < 51	0.20	Half or more of the lot is in the FEMA regulatory floodway
Share of lot in the FEMA regulatory floodway score < 100	0.60	Part of the lot is in the FEMA regulatory floodway
Share of lot over mapped mines (City undermined area) score < 51	0.90	Over mapped mines: anything larger than a single house is prohibited until a mine investigation clears it (§906.05)
Lot area (for the sliver-lot check) score < 100 AND Site availability (what is on the parcel now) ≠ 3	0.30	Sliver lot under 600 sq ft

§05 · housing types

Housing types: legal pathway and physical fit

The bottom panel scores 16 housing types on a parcel in two independent layers: a policy legal pathway read from the City zoning code, and a model physical site fit. They are shown side by side and never blended into one number, so a lot that would physically fit a duplex but isn't zoned for one reads as a rezoning question, not a dead end. Both feed the verdict (§6) as separate pass/fail reasons.

E9 · Legal pathway score (tile number)

T = ⎩ ⎨ ⎧ 100856040 round (5 + 30 c ℓ) by right administrator exception special exception conditional use not permitted

$c$: rezoning closeness: 1 if a district in the same family allows it (density-suffix change), 0.5 if one step up/down the R1D → R1A → R2 → R3 → RM ladder, 0.2 if only a distant district, 0 if none
$ℓ$: likelihood: Zoning Board relief approval rate in this base district, 2023–26 (all case types); 0.7 where there are no local cases

Not-permitted tops out at 35, below conditional use (40): it always means extra process, but it isn't a flat zero either. Planned-unit districts and Mount Oliver get no single number.

E10 · Jev site fit (typed rubric)

E label = i = 0 \sum L - 1 i p_{i} fit = \frac{E}{L - 1} = level_{round (E)} review if conf < 0.3 (House: 0.2)

$L$: 4 levels: 0 Cannot fit · 1 Fits only with major compromises · 2 Fits with minor compromises · 3 Comfortable fit
$p_{i}$: the model's probability for each level
$facts$: lot area, bounding-rectangle width × depth, district minimum lot size, hazard shares ≥ 1%, all computed in PostGIS/code first

The expected value, not the model's top pick: a near 50/50 split between two levels lands between them instead of snapping to a falsely confident extreme.

Approval pathways

8 pathways · Zoning Code Ch. 922

Pathway	Who decides	Hearing	Clock	If the City misses it	§
By right	Zoning staff (BDA review via OneStopPGH)	no	none	n/a
Administrator exception	Zoning Administrator (Administrator Exception)	no	decision 21 days after complete application	City code: deemed denial, may go to ZBA (§922.08.C)	§922.08
Special exception (Zoning Board hearing)	Zoning Board of Adjustment	yes, >=21-day notice	hearing within 45 days of complete application; decision within 45 days of hearing	conflict: City code says deemed denial; PA MPC (2003 ed.) says deemed approval; see sources/pa-dced-2003-01-mpc-908-913-2-deemed-approval.md §908(9)	§922.07
Conditional use (Planning Commission + Council)	Planning Commission recommends, City Council decides	yes, two (PC and Council)	up to 4 x 45 days (~6 months) before deemed denial	conflict: City code says deemed denial; PA MPC (2003 ed.) says deemed approval; see sources/pa-dced-2003-01-mpc-908-913-2-deemed-approval.md §913.2(b)(2); 39 Council CUs recorded as Passed pursuant to Case Law	§922.06
Not permitted (variance or rezoning only)	Only via use variance (ZBA, §922.09 five findings incl. hardship) or rezoning (Council map amendment)	yes	variance as ZBA; rezoning not bounded	variance: conflict: City code says deemed denial; PA MPC (2003 ed.) says deemed approval; see sources/pa-dced-2003-01-mpc-908-913-2-deemed-approval.md §908(9)	§922.09 / map amendment
Set by the site's approved plan (planned-unit district)	Planning Commission, through the site's approved unit development plan	yes	not bounded in code read	unknown	§909.02
Mount Oliver Borough (not City zoning)	Mount Oliver Borough
unknown

Typology × district matrix

16 types × 57 districts · as of 2026-09-26 · §911.02

EMI

GPRA

GPRB

GPRC

GT-A

GT-B

GT-C

GT-D

GT-E

LNC

MTOBOR

NDI

NDO

R-MU

R1A-H

R1A-M

R1A-VH

R1D-H

R1D-L

R1D-M

R1D-VH

R1D-VL

R2-H

R2-L

R2-M

R2-VH

R3-L

R3-M

RIV-GI

RIV-IMU

RIV-MU

RIV-NS

RIV-RM

RM-H

RM-M

RM-VH

RM-VL

SP-1

SP-10

SP-11

SP-4

SP-5

SP-8

SP-9

UC-E

UC-MU

UNC

UPR-A

UPR-B

Single-unit detached house

Single-unit attached (rowhouse)

Two-unit (duplex)

Three-unit

Multi-unit apartments (4+)

Housing for the elderly (limited, <30 units)

Housing for the elderly (general, 30+ units)

Assisted living A (<9 beds)

Assisted living B (9–17 beds)

Assisted living C (18+ beds)

Personal care residence (small)

Personal care residence (large)

Community home

Multi-suite residential (limited)

Multi-suite residential (general)

Interim housing

By rightAdministrator exceptionSpecial exception (Zoning Board hearing)Conditional use (Planning Commission + Council)Not permitted (variance or rezoning only)Set by the site's approved plan (planned-unit district)Mount Oliver Borough (not City zoning)

Zoning Board outcomes by base district (the likelihood term in E9)

ZBA decision PDFs, 2023–26

Base district	Cases	Approved	Denied	Approval rate
R1D	101	87	11	86%
R1A	83	74	7	89%
R2	62	46	13	74%
LNC	49	45	3	92%
RM	31	30	1	97%
UI	23	20	2	87%
UNC	17	12	3	71%
EMI	13	11	1	85%
H	12	10	1	83%

Only posted decisions are counted, and the rate covers all relief types, not rezonings specifically. It is an approximation of how a district treats requests the code doesn't allow outright.

§06 · verdict

Can it be built? The verdict and the pencil check

A second answer beside the scores. The pillars rank how good a place is; the verdict says what stands in the way, per housing type, from code pass/fail checks. The worst reason decides the color. The framing came from a hackathon housing expert: red = not developable, yellow = could be developable with variances or subsidy, green = developable as-is.

Color	Check	When
red	Zoning	Not permitted and no district of similar density allows it
yellow	Zoning	Special exception or conditional use; or not permitted but a nearby-density district allows it (rezoning or use variance). Each reason carries its approval clock.
green	Zoning	By right or Zoning Administrator approval
unknown	Zoning	District not in the use table, planned district, or Mount Oliver
red	Site	Half or more floodway; sliver lot under 600 sq ft; park, cemetery, rail or right-of-way
yellow	Site	Part floodway; mostly floodplain, 25%+ slope or landslide-prone; recent landslide; over mapped mines (investigation or cover check); condo unit, large building or institution
yellow	Width	Lot width minus §903.03 interior side setbacks leaves less than the minimum building width (likely dimensional variance)
red / yellow	Fit	Jev rates 'Cannot fit' (red) or below the halfway point (yellow)
red / yellow / green	Pencil	Value per unit vs. cost per unit with margin (below)
unknown	Data	Missing hazard, sale or rent data. Unknown outranks green, so missing data never reads as buildable.

Side-setback check (§903.03)

min building width 14 ft

Total interior side setback (both sides, ft) by base district and density suffix. Buildable width = lot width minus this. Attached (party-wall) housing has no interior side setback, so it is skipped. Corner-lot exterior side setbacks and contextual setbacks (§925.06) are not modeled. Lot width is the short side of the lot's oriented bounding rectangle, so irregular lots are approximate. Assumption: about the narrowest practical detached building (a narrow Pittsburgh rowhouse is ~14–16 ft). Below this, a hackathon SME's example (24 ft RM-M lot, two 10 ft setbacks, 4 ft left) says a dimensional variance is likely granted but costs time and money. Source: Pittsburgh Code §903.03 as amended Ord. 10-2025 (eff. 2025-05-07), read on eCode360 2026-09-26: research/sources/ecode360-2026-09-26-pittsburgh-903-03-dimensional-standards.md

Base district	VL	L	M	H	VH
R1D	15 ft	10 ft	10 ft	10 ft	10 ft
R2	15 ft	10 ft	10 ft	10 ft	10 ft
R3	15 ft	10 ft	10 ft	10 ft	10 ft
R1A	10 ft	10 ft	10 ft	10 ft	10 ft
RM	60 ft	50 ft	20 ft	20 ft	20 ft

E11 · Pencil check (per unit)

cost value status = sf \cdot c_{sf} (1 + soft) + \frac{site + steep}{units} need = cost (1 + margin) = {tract median sale price 103 \times monthly rent houses 2+ units = ⎩ ⎨ ⎧ pencils tight subsidy no value \geq need value \geq need at c_{sf} = 150 value \geq 0.35 need otherwise

$c_{sf}$: construction $/sf, editable; presets 150 / 225 / 350
$site$: $37,500 per building, editable
$steep$: $30,000 when half the lot is 25%+ slope
$soft$: 15% of hard cost
$margin$: 10%

Does it pencil? A first screen of value against cost per unit, the check hackathon SMEs said developers run before zoning or variances (2026-09-27): 'cost exceeding market value is the biggest challenge.' Value is what nearby homes sell for (close to what an appraiser's comparable sales would show) or what nearby rents capitalize to; cost is construction plus site work plus soft costs and a margin. Land cost, financing terms and subsidies are not modeled. Every number below is editable in the panel and is a value judgment or a practitioner estimate, not a quote.

Assumption	Value	Source
$/sf, production builder	$150	Hackathon SME (Housing Innovation Alliance), 2026-09-27: high-volume production builders would imply about $150/sf if they entered the City market. Matches NAHB's 2024 national cost survey without its site-work stage ($149/sf; Pittsburgh's RSMeans residential location factor is 1.01).
$/sf, typical infill (default)	$225	Hackathon SME (Housing Innovation Alliance), 2026-09-27: $200–250/sf for single-family infill in the City, varying with builder scale.
$/sf, small builder	$350	Hackathon SME (practitioner), 2026-09-27: $325–375/sf vertical construction, excluding site work; no reliable local public source.
Site work per building	$37,500	Hackathon SME (practitioner), 2026-09-27: $25k–50k for a single-unit City lot (water and sewer taps, grading, sidewalks, landscaping); midpoint. Depends on utility depth, soil, and demolition debris in old basements.
Steep-lot adder	$30,000	Assumption: the low end of a builder blog's $30k–80k hillside site work (research/knowledge/methods/pro-forma.md); applied when half or more of the lot is 25%+ slope.
Soft costs	15%	Assumption: design, permits, fees and carrying costs as a share of hard cost; SMEs named soft costs but gave no figure.
Margin	10%	Assumption: value must beat cost by a margin 'large enough to attract investors and get banks to lend' (hackathon SME, Pro-Housing Pittsburgh); 10% is a placeholder.
Rent → value	× 103	Assumption: value per unit = monthly rent × 12 × (1 − 40% operating costs) ÷ 7% cap rate ≈ 103 × monthly rent.
Subsidy floor	35%	Value judgment: if value covers at least 35% of the cost (with margin) at the user's assumption, call it a subsidy candidate (yellow); otherwise it does not pencil (red). URA's per-unit caps are shown beside the gap for scale; any gap within a cap already clears this floor at realistic costs. At the defaults, red covers about 16% of vacant or occupied parcels for a house, 36% for a duplex and 11% for apartments. A stricter rule (red whenever the gap exceeds URA's cap) would make about 63% of houses and 88% of duplexes red; we did not use it because the value inputs understate new-build value and URA's Rental Gap Program needs 4+ units.
URA gap caps	$130,000 sale / $75,000 rent	Shown next to every gap for scale. For-sale: URA For-Sale Development Program, max $130,000 per newly constructed unit, buyers at or below 80% AMI. Rental: URA Rental Gap Program (guidelines amended Aug 2024), max $75,000 per unit at 30% AMI ($50k at 50%, $35k at 60%); URA often waives its caps, and other sources (LIHTC, PHFA) can stack. Read 2026-09-27: research/sources/ura-2026-09-27-rgp-fsdp-guidelines-and-hof-2026-plan.md. How much gap financing a project can get 'can vary wildly' (SME).
Unit sizes	single_detached 1×1400 sf, single_attached 1×1400 sf, two_unit 2×1100 sf, three_unit 3×1000 sf, multi_unit 12×850 sf	Unit sizes are assumptions; 1,400 sf is the SME example house. Houses are valued as for-sale homes; two or more units as rentals.

Values come from existing homes (tract resale prices) and existing rents (ACS block groups, which are noisy: 48 of 265 City block groups have a CV above 30%). Both understate what new construction sells or rents for. City new-build sales run about $370–390/sf, against about $180/sf for all valid sales, roughly 2×, but only 15 of 90 neighborhoods have 5+ new-build comps in 5 years (research/sweeps/r9-revenue-comps-sub-zip.md). No new-build premium is applied, so the check leans toward 'doesn't pencil'. Read it as the appraisal-gap risk, not a verdict on the project.

Permit valuations are not a usable substitute: 220 new 1–2 family City permits (2019–24, deduplicated, assessed living area built after the permit) give a median $104/sf, and the declared value was a median 0.31× the same homes' later sale price (n=63), so permit values understate cost. Reproduce with research/admin/scripts/permit_cost_per_sf.py (research/sweeps/r9-permit-cost-per-sf.md).

Not priced: land, landslide geotechnical work, floodplain elevation, mine investigation, environmental remediation, financing.

How long each approval can take before a building permit, shown with the pathway reason so hearings read as time and carrying cost, not just a probability (hackathon SME, 2026-09-27: variances are usually granted, but the time and expense make lots hard to develop). Code deadlines are maximums from the City code; the rezoning figure is observed.

§07 · ai

The AI components and their guardrails

Each model gets narrow, pre-computed inputs and has a documented failure mode. No model output becomes a score without being labeled as such.

Jev site-fit modelmodel

A System One decision API (TypeSafe's Jev, served via OpenRouter). It returns typed answers with probabilities. There is no free text to parse or hallucinate.
One score question per housing type (House, Townhouse, Duplex, Triplex, Apartment, Elderly housing, Assisted living (small), Assisted living (mid), Assisted living (large), Personal care (small), Personal care (large), Community home, Multi-suite (limited), Multi-suite (general), Interim housing), asked in a single call against the same facts.
Facts are computed first in PostGIS and code: lot area, oriented-rectangle width and depth, district minimum lot size, hazard shares. The model does no arithmetic.
Every type is rated, including ones zoning forbids, because physical fit is what makes a rezoning ask worth pursuing.
Your pillar weights are not sent: a physical judgment shouldn't move with preferences.
Answers are cached per (model, facts, types). On timeout (10 s) or a missing key the panel says the fit is unavailable. It never shows a guessed number.

Chat companionllm

Google Gemini: gemini-3.5-flash-lite → gemini-3.1-flash-lite → gemini-2.5-flash-lite. If a model errors, or hasn't answered after 4 s, the next one starts too and the first answer wins; each call times out at 12 s.
Sees only what the panels show for the selected parcel (scores, breakdowns, typology tiles with Jev's site fit, alerts, zoning) plus standing definitions. Each fact has an id, source and as-of date. Without a parcel, it only explains how the tool works.
Replies must cite fact ids. Citations to ids that weren't provided are removed.
Number guard: any sentence containing a number that no provided fact states is dropped. Small counting numbers (0–10) and a fraction's percent form (0.81 → 81) are allowed.
What-ifs are computed, never estimated: the rescore tool reruns E5–E6 with new weights or a preset, and rezoning and vacant-land scenarios are scored by the real scorer ahead of time. Anything else gets "not computed".
At most 10 turns of history and 3 tool steps per answer. Without a key, it shows the screen's own notes instead.

§08 · map layers

Every map layer and where it comes from

Read live from the overlay registry (components/map/overlays). Every layer has to declare its source, date, geography and evidence type before it can be added, and the same metadata appears in the Layers pane.

Layer	Group	Evidence	Source	As of	One value describes	Caveats
License plate readers (ALPR) alpr-cameras	cameras	observed	OpenStreetMap contributors (ODbL), incl. DeFlock mapping deflock.me	OSM snapshot pulled 2026-09-26	Mapped camera points, positions as placed by volunteer mappers	3 caveats Crowd-mapped, not an official inventory: coverage is incomplete and some entries may be stale. Most points have no operator tag, so police vs. retail use can't be told apart for them. About 310 points tagged PlateSmart/Cyclops look bulk-mapped; treat those with caution.
Live cameras cameras	cameras	observed	511PA (PennDOT, PA Turnpike), USGS HIVIS, CMU Breathe Cam, WeatherSTEM, EarthCam, PixCams, Pittsburgh Zoo 511pa.com	Feeds are live; camera list built 2026-09-26	Camera locations; some webcam positions are approximate (noted in the viewer)	2 caveats Only feeds their owners publish. Cameras go offline without notice. 511PA traffic cameras show a still image refreshed about every 10 seconds, not video.
Condemned properties condemned-properties	development	observed	City of Pittsburgh condemned and dead-end properties (WPRDC, CC-BY) data.wprdc.org	Updated daily; pulled Sep 2026	Property locations	2 caveats City of Pittsburgh only; no county-wide open data exists. Condemnation does not establish demolition feasibility or site availability.
New homes & demolitions (permits) permits-activity	development	observed	City of Pittsburgh PLI permits (WPRDC, CC-BY) data.wprdc.org	Issued Jan 2022 – Sep 2026	Permit locations	2 caveats City of Pittsburgh only; no county-wide open data exists. Issued is not built; permits carry no unit counts.
Impervious surface impervious	environment	observed	MRLC National Land Cover Database percent developed impervious 2021 (public domain) mrlc.gov	2021	30 m grid	1 caveats A stormwater and heat proxy; parcel-level billed impervious area is kept by PWSA.
Surface heat (2020) surface-heat	environment	observed	City of Pittsburgh urban heat island layer: 2020 mean land surface temperature tiles.arcgis.com	2020 composite	Satellite grid covering the county	2 caveats Surface (roof and pavement) temperature, not air temperature. Relative only: the tiles are a grayscale stretch, so absolute temperatures aren't shown.
Tree canopy tree-canopy	environment	observed	USFS / MRLC National Land Cover Database tree canopy cover 2021 (public domain) mrlc.gov	2021	30 m grid	1 caveats 30 m resolution misses street trees and small yards.
City landslide-prone & undermined overlays city-hazard-overlays	hazard	policy	City of Pittsburgh zoning hazard overlays (Zoning Code Ch. 906), City ArcGIS services1.arcgis.com	Current City layers (pulled Sep 2026)	City of Pittsburgh only	1 caveats Regulatory overlays, not site surveys. The county landslide and DEP mined-out layers cover more ground.
City steep-slope overlay (25%+) city-steep-slopes	hazard	policy	City of Pittsburgh zoning hazard overlays (Zoning Code Ch. 906), City ArcGIS services1.arcgis.com	Current City layers (pulled Sep 2026)	City of Pittsburgh only	2 caveats Loads at zoom 14 and closer. This is the City's regulatory layer; the USGS slope raster covers the whole county.
Flood zones flood-zones	hazard	observed	FEMA National Flood Hazard Layer (effective FIRMs) hazards.fema.gov	Effective FIRM panels (dates vary by panel)	Mapped flood zone polygons; check parcel overlap, not just its center	3 caveats FIRMs map riverine flooding only, not flash floods, basement backups or combined-sewer flooding. Some panels are decades old. Not an insurance or legal flood determination.
Landslide damage sites (FEMA aid) landslide-public-assistance	hazard	observed	Allegheny County landslide public-assistance sites (County ArcGIS) services1.arcgis.com	Pulled Sep 2026 (event dates not published in the layer)	Damage site points	2 caveats Public infrastructure damage only; private property slides aren't included. Damage values are the layer's estimates.
Landslide incidents (county roads) landslide-incidents	hazard	observed	Allegheny County Public Works landslide records services1.arcgis.com	Edited Sep 2025	Incident points on county-maintained roads	2 caveats County roads only; city and private slides are not included. Some dates are missing.
Landslide-prone areas landslide-susceptibility	hazard	observed	USGS-era landslide susceptibility mapping (1970s–80s), via Allegheny County GIS services1.arcgis.com	County layer edited Jul 2018; underlying mapping decades old	Mapped susceptibility polygons; check parcel overlap, not just its center	2 caveats Provenance inferred: the county service has no description or license. Susceptibility, not a prediction. Site-specific geotechnical review is still required.
Mine subsidence risk zone mine-subsidence	hazard	observed	PA DEP Mine Subsidence Insurance, subsidence risk (mining confirmed) gis.dep.pa.gov	PA DEP MSI service	Broad risk zones	2 caveats Coarse zones for insurance outreach, not parcel-level findings. See the PA Mine Subsidence Insurance program.
Serious traffic crashes serious-crashes	hazard	observed	PennDOT reportable crashes, Allegheny County (WPRDC, CC0) data.wprdc.org	2023–2025 (2025 file updated Sep 2026)	Crash locations	2 caveats Reportable crashes only. A few crashes without coordinates are not shown.
Steep slopes slope	hazard	observed	USGS 3D Elevation Program (3DEP), slope computed on the fly elevation.nationalmap.gov	3DEP service dated 2026-08-25	~1–10 m elevation grid, resampled to the zoom level	2 caveats Bare-earth terrain: retaining walls and graded yards are not reflected. The 25% line is the zoning threshold, not a geotechnical finding.
Tornado paths (1950–2025) tornado-paths	hazard	observed	NOAA Storm Prediction Center tornado paths (public domain) spc.noaa.gov	1950–2025	Recorded tracks (straight start→end lines for older records)	2 caveats History, not a probability map; see the weather-risk layer for modeled tornado risk. F-scale before 2007, EF-scale after.
Undermined areas (coal mines) mined-out-areas	hazard	observed	PA DEP mined-out areas, underground coal gis.dep.pa.gov	PA DEP District Mining Operations layer	Mine footprints compiled from historic mine maps	2 caveats Boundaries are approximate; not being mapped here does not mean not mined. Undermined land can require a site investigation before building (city §906.05).
Air quality air-quality	heat	observed	EPA EJScreen v2.32, via the Public Environmental Data Partners mirror (EPA took EJScreen offline in 2025) services2.arcgis.com	EJScreen v2.32; indicator years vary	Census block group average, not this parcel	2 caveats Modeled estimates, not local monitor readings. Which pollutant matters most is a value judgment.
Car dependence & housing cost (LAI) location-affordability	heat	observed	HUD / DOT Location Affordability Index v3 services.arcgis.com	LAI v3 (built on 2012–2016 ACS inputs; HUD's latest release)	Census tract (2010 boundaries)	3 caveats Old inputs (2012–2016): relative differences between tracts are the useful part. H+T share: HUD's LAI sets no affordability threshold; the Center for Neighborhood Technology's H+T Index uses 45% of income. It's modeled for a fixed household, so tracts differ only by location. Miles driven is HUD's modeled VMT for a fixed household profile (median income, 4 people, 2 commuters), so tracts differ only by location. HUD's observed VMT covers Illinois only.
Children's elevated blood lead child-blood-lead	heat	observed	Allegheny County Health Department, Elevated Blood Lead Level Rates (WPRDC, CC0) data.wprdc.org	Pooled 2015–2020 (dataset updated May 2026)	Census tract (2010 boundaries)	4 caveats Share of children tested, not all children; testing has been mandatory for under-6s since 2018. Venous-confirmed ≥ 5 µg/dL (ACHD's documented definition; CDC's reference level has been 3.5 since 2021). ACHD's newer 2021–2024 columns aren't shown: their definition isn't published and they run 5–10× ACHD's reported county rate (~3.1% in 2023). Faded tracts had fewer than 10 elevated results. Pre-1978 paint is the main local source, not only water lines.
Code violations (last 24 months) code-violations	heat	observed	City of Pittsburgh PLI/DOMI/ES violations (WPRDC, CC0); housing units from ACS 2024 5-year data.wprdc.org	Last 24 months, pulled Sep 2026	Census block group rate (city only)	3 caveats City of Pittsburgh only; no county-wide open data exists. Complaint-driven: reflects reporting behavior as well as conditions. Includes refuse and weeds violations, not only building conditions.
Economic mobility (Opportunity Atlas) opportunity-atlas	heat	observed	Opportunity Insights / U.S. Census Bureau, The Opportunity Atlas opportunityinsights.org	Children born 1978–83, measured as adults	Census tract (2010 boundaries)	2 caveats Historical cohort outcomes, not current neighborhood conditions. Faded tracts have a standard error over 5 percentile points.
Eviction filings 2025 evictions	heat	observed	Princeton Eviction Lab, Eviction Tracking System (Pittsburgh); renter households from ACS 2024 5-year evictionlab.org	Filings Jan–Dec 2025	ZIP code (ZCTA), not this parcel	3 caveats Filings, not completed evictions. Cases with a missing or bad ZIP (153 in 2025) aren't on the map, so ZIP counts are undercounts. PO-box ZIPs have no area and aren't drawn; a few ZIPs extend past the county line.
Home internet (ACS) home-internet	heat	observed	U.S. Census Bureau ACS 2024 5-year, table B28002 (via Census Reporter) censusreporter.org	ACS 2020–2024	Census tract estimate (2020 boundaries), not this parcel	2 caveats Job searches, school, telehealth and housing applications (including subsidized waitlists) are mostly online. Tract margins of error are large; the tooltip shows ± in percentage points.
Housing cost burden (CHAS) chas-cost-burden	heat	observed	HUD Comprehensive Housing Affordability Strategy (CHAS) 2018–2022, Table 8 huduser.gov	CHAS 2018–2022 (released Dec 2025)	Census tract share (2020 boundaries), not this parcel	3 caveats Five-year ACS-based tabulation with large tract margins of error. Income bands use HUD Area Median Family Income (HAMFI), which differs slightly from HUD income limits. Counts are rounded by HUD.
Housing voucher use housing-vouchers	heat	observed	HUD Housing Choice Vouchers by Tract services.arcgis.com	Vouchers through Dec 2025	Census tract share (2020 boundaries), not this parcel	2 caveats Where voucher holders live, which reflects landlord acceptance as well as demand. HUD omits tracts with 10 or fewer voucher holders for privacy (200 of 394 here), so grey means few vouchers, not missing data.
Income & housing cost housing-costs	heat	observed	U.S. Census Bureau, ACS 2024 5-year (2020–2024), via Census Reporter censusreporter.org	ACS 2024 5-year release	Census block group estimate, not this parcel	3 caveats Five-year average, not current conditions. Faded areas have a margin of error over 40% of the estimate. The ACS vacancy rate includes normal turnover; "other" vacant is the closest proxy for abandonment.
Jobs & transit access jobs	heat	observed	U.S. Census Bureau LEHD LODES8 WAC 2023 (public domain); UMN Accessibility Observatory, Access Across America: Transit 2024 (CC BY-NC 4.0) lehd.ces.census.gov	Jobs 2023; transit access 2024	Census block group total, not this parcel	4 caveats Jobs are counted at the employer's reporting location; some multi-site employers report at headquarters. Transit access is a modeled weekday-morning average. UMN data is licensed for non-commercial use only. Job change compares 2019 and 2023 LODES; a single employer moving can swing a block group.
Legal pathway by housing type legal-pathway	heat	policy	City of Pittsburgh Zoning Code §911.02 use table, transcribed by the research team; zoning from PGHWebZoning ecode360.com	Code as of 2026-09-26	City of Pittsburgh zoning district (city only; suburbs have their own codes)	6 caveats Working research, not legal advice. Zoning Board counts cover posted decisions only (withdrawn cases have none), so approval rates skew high. Faint districts are unconfirmed readings or inferred from a district's adopted use list; the tooltip explains why. The use table is not the only gate: dimensional standards, overlays, Site Plan Review at 4+ units and historic review add steps. Housing for the elderly limited and general have different permissions; pick the one that matches the project size. Pending Bills 2025-1545 (ADUs, parking) and 2026-0834 (Ch. 922 procedures) would change some cells.
Life expectancy (2010–15) life-expectancy	heat	observed	CDC / NCHS U.S. Small-Area Life Expectancy Estimates Project (USALEEP) cdc.gov	Deaths 2010–2015	Census tract (2010 boundaries)	2 caveats Faded tracts have a standard error over 2 years. Older data; conditions may have changed.
Market type & displacement risk market-mva	heat	observed	Reinvestment Fund Market Value Analysis 2021 and Displacement Risk Ratio, for URA / Allegheny County (WPRDC, CC0) data.wprdc.org	MVA 2021 (2017–19 sales); DRR through 2019/20	Census block group (2010 boundaries)	3 caveats Pre-pandemic data; the market has moved since. The DRR measures price growth relative to what long-time residents can afford; its exact formula is not fully documented. DRR classes are the source's own; 579 of 1,114 block groups share the lowest class, so the layer mainly highlights the higher-risk minority. Data ends 2019/20.
Neighborhood health (CDC PLACES) health-outcomes	heat	observed	CDC PLACES, census tract estimates (2024 release) cdc.gov	PLACES 2024 release (mostly 2022 BRFSS)	Census tract estimate (2020 boundaries), not this parcel	2 caveats Model-based estimates, not local surveys. No local estimate of housing insecurity, utility shutoffs or food insecurity exists for Pennsylvania (PA didn't field that survey module).
Prices & rents by ZIP market-zip	heat	observed	Allegheny County property sales (WPRDC, CC0); Zillow Observed Rent Index (Zillow Research) zillow.com	Valid sales Jan 2024 onward; ZORI through Aug 2026	ZIP code (ZCTA); not neighborhoods	2 caveats County sales need 1–2 months to be validated, so the latest months are missing. ZORI tracks listed asking rents, not what existing tenants pay.
Safety (reported) safety	heat	observed	City of Pittsburgh police blotter, UCR Part I (WPRDC); Allegheny County homicide incidents (ACHD / Medical Examiner); population from ACS 2024 5-year data.wprdc.org	City crime Dec 2020 – Nov 2023 (blotter ends); homicides 2016–2022 and 2024 (no 2023 layer)	Census tract rate (2020 boundaries), not this parcel	3 caveats Reported incidents reflect reporting and police deployment, not the risk people experience. Suburbs have no incident-level crime data; only homicides cover the whole county. Only tract rates are shipped; no individual incidents or victim details.
Social vulnerability & child opportunity equity	heat	observed	CDC/ATSDR Social Vulnerability Index 2022 (public domain); Child Opportunity Index 3.0, diversitydatakids.org atsdr.cdc.gov	SVI 2022; COI 2021 data year	Census tract rank (2020 boundaries), not this parcel	3 caveats SVI ranks are national percentiles; COI ranks are within the metro. COI shows gaps in neighborhood investment, not whether a place is a 'good place to live'. COI comes from an unofficial ArcGIS mirror of the official release.
Vacant addresses (USPS) usps-vacancy	heat	observed	HUD/USPS vacant address data (WPRDC, CC0) data.wprdc.org	2023 Q4	Census tract share (2020 boundaries), not this parcel	2 caveats Vacant means the carrier reports mail uncollected for 90+ days. Students and seasonal residents can inflate it.
Walkability (EPA) walkability	heat	observed	U.S. EPA National Walkability Index (Smart Location Database) epa.gov	Inputs circa 2018–2019	Census block group (2020)	2 caveats Ignores Pittsburgh's hills, public steps and sidewalk gaps. Transit distance is from EPA's 2018-era GTFS, not current Port Authority service.
Weather risk weather-risk	heat	observed	FEMA National Risk Index services.arcgis.com	NRI December 2025	Census tract rating (2020 boundaries), not this parcel	3 caveats NRI risk includes the dollar value of exposed buildings, so dense or valuable tracts rate higher. Use the flood-zone, landslide and City hazard-overlay layers for parcel-precise hazards. Some hazards barely vary here: heat wave, for example, is only rated Very Low or Relatively Low in every tract, so its colors don't signal a real difference.
Zoning: residential density allowed residential-zoning	heat	policy	PGHWebZoning (City of Pittsburgh), transcribed by the research team ecode360.com	2026-09-26	City zoning district (city only; Mount Oliver Borough uses its own code)	2 caveats Colored by the base district's typical density tier only; the actual permitted-use table has exceptions per typology -- see the Legal pathway overlay for exact rules. Working research, not legal advice.
Lead service lines lead-service-lines	infrastructure	observed	Pittsburgh Water (PWSA) service line material services5.arcgis.com	PWSA layer, edited Sep 2026	One point per service address (PWSA service area only)	4 caveats Non-lead addresses (~61k of ~81k) are not drawn. Covers PWSA customers only (most of the City). Outside PWSA, no dot means no data, not no lead. Unknown means not yet inspected, not safe. PWSA publishes this with an accuracy disclaimer.
Sewer lines sewer-lines	infrastructure	observed	Allegheny County sewer lines (3 Rivers Wet Weather / LBs export) services1.arcgis.com	2012 export (last edited Jan 2018)	Pipe segments; loads for the visible area at zoom 15+	2 caveats Dated snapshot: newer PWSA work is not reflected. Capacity is not public, so this shows where pipes are, not whether they can take new load.
Transit stops transit-stops	infrastructure	observed	Pittsburgh Regional Transit stops (WPRDC, CC-BY) data.wprdc.org	Updated 2026-09-22 (GTFS feed 2606, service from 2026-06-28)	Stop points; heatmap below zoom 12 is weighted by weekday trips	2 caveats Scheduled trips, not realized reliability. Walking distance to a stop is longer than it looks on hilly terrain.
City-owned property city-owned-land	land	observed	City of Pittsburgh city-owned property (WPRDC, CC-BY) data.wprdc.org	Updated daily; pulled Sep 2026	Parcel locations	2 caveats City of Pittsburgh only. Available does not mean buildable; check lot size, slope and zoning.
Long-term tax delinquency tax-delinquent	land	observed	City of Pittsburgh and school district property tax delinquency (WPRDC, CC-BY) data.wprdc.org	Pulled Sep 2026	Parcel locations, shown from zoom 14	3 caveats City of Pittsburgh only. These are real households; no owner information is shown. The 3-year threshold is an assumption, not an official definition.
Upcoming tax sales treasury-sales	land	observed	City of Pittsburgh treasurer sales (WPRDC; license not specified) data.wprdc.org	Next sale Oct 2, 2026	Property locations	2 caveats City of Pittsburgh only. Properties can be withdrawn before the sale.
Vacant buildings (USPS, City) vacant-buildings	land	observed	City of Pittsburgh, USPS vacancy flags by parcel (Vacant_USPS_Feb_24), joined to City parcel records services1.arcgis.com	USPS snapshot Feb 2024	Parcel centroids, City of Pittsburgh only	2 caveats USPS 'vacant' means no mail collected for 90+ days; it includes some renovations and seasonal cases. A 2024 snapshot; some buildings have since been reoccupied or demolished.
Vacant lots vacant-lots	land	observed	Allegheny County Office of Property Assessments via WPRDC (vacant-land use codes) data.wprdc.org	Assessments as of Sep 2026	Parcel centroids, all of Allegheny County	3 caveats The assessor's land-use code, not a site visit; some 'vacant' parcels are parking, yards or wooded slopes. Most vacant lots have no house number, only a street name. Owner type is the assessor's category; no owner names are included.
800 ft care-facility spacing (partial) care-facility-spacing	legal	policy	Research team, from PA DHS licensed personal care / assisted living and CMS nursing homes; rule from Zoning Code §911.04 ecode360.com	Facilities as of 2026-09-26	800 ft circles, City of Pittsburgh	2 caveats Partial: only licensed facilities are drawn; unlicensed group homes also count under the rule but aren't mapped (privacy). A parcel-level gate the district map can't show.
City Council land-use votes (2000–26) council-land-use-actions	legal	observed	City of Pittsburgh Legistar (City Council), compiled and geocoded by the research team pittsburgh.legistar.com	2000 – Sep 2026, pulled 2026-09-26	Action locations, City of Pittsburgh	4 caveats 101 of 284 matters had no mappable address and aren't shown. Area-wide remaps are shown as a single point. Whether a change adds or removes housing is our coding of the district change, not the City's. Non-residential actions are hidden.
Senior & supportive housing (existing) senior-supportive-housing	legal	observed	PA DHS provider directory, PA DOH / CMS nursing homes, HUD Multifamily, HUD LIHTC, HACP (compiled by the research team) humanservices.dhs.pa.gov	Pulled 2026-09-26	Facility locations, City of Pittsburgh only	4 caveats One building can appear under HUD, LIHTC and HACP (124 points ≈ 101 sites); don't add points up. The zoning use is our inference from size and program, not a City determination. Many sit where their use isn't permitted today; most likely they predate the current code. Market-rate senior apartments aren't covered; group homes are left out for privacy.
Where each housing type got permitted (2019–26) permits-by-type	legal	observed	City of Pittsburgh OneStopPGH permits and Development & Construction Projects (compiled by the research team) data.wprdc.org	Permits May 2019 – Sep 2026, pulled 2026-09-26	Permit locations, City of Pittsburgh only	4 caveats Housing type is classified from permit text by keyword; 129 unclassified permits are hidden. Zoning is today's district, which may differ from the one at permit time. One project can have several permits; don't count points as projects. Dot size = units where known.
Zoning Board decisions: housing (2023–26) zba-decisions-housing	legal	observed	City of Pittsburgh ZBA decisions (pittsburghpa.gov and Internet Archive), coded by HouseHack 2026-09-26 pittsburghpa.gov	Decisions 2023 – Aug 2026, pulled 2026-09-26	Case addresses, City of Pittsburgh	4 caveats Covers roughly 61–73% of each year's case numbers; withdrawn cases never get a posted decision, so approval looks higher than it is. Zoning is the district named in each decision; some name several. Each case's decision PDF is in the data (decision_pdf); popups can't hold links. Outcomes were read from the decision text (some scanned and OCR'd).
Zoning Board decisions: other (2023–26) zba-decisions-other	legal	observed	City of Pittsburgh ZBA decisions (pittsburghpa.gov and Internet Archive), coded by HouseHack 2026-09-26 pittsburghpa.gov	Decisions 2023 – Aug 2026, pulled 2026-09-26	Case addresses, City of Pittsburgh	4 caveats Covers roughly 61–73% of each year's case numbers; withdrawn cases never get a posted decision, so approval looks higher than it is. Zoning is the district named in each decision; some name several. Each case's decision PDF is in the data (decision_pdf); popups can't hold links. Outcomes were read from the decision text (some scanned and OCR'd).
Access to Opportunity pillar-access	pillar	value	HouseHack pillar scores (pillars.config.json, default weights) github.com	2026-09-27.1	Distance-weighted mean of nearby City of Pittsburgh parcels, drawn as hexagons (~170 m across, ~85 m when zoomed in), not one parcel	5 caveats 100 = a good place to build new housing. Scores combine observed data using weights that are value judgments. Colors are relative: the bins are City hex percentiles (p10–p90), not fixed score bands. Smoothed: each hex blends parcels around it (Gaussian, widening until it reaches 8 scored parcels), so sharp edges such as a floodway bleed into neighboring hexes. A hex mean can hide a capped parcel; the tooltip counts those inside the hex. Many inputs are tract or block-group values, so neighboring hexes can share them.
Affordability pillar-afford	pillar	value	HouseHack pillar scores (pillars.config.json, default weights) github.com	2026-09-27.1	Distance-weighted mean of nearby City of Pittsburgh parcels, drawn as hexagons (~170 m across, ~85 m when zoomed in), not one parcel	5 caveats 100 = a good place to build new housing. Scores combine observed data using weights that are value judgments. Colors are relative: the bins are City hex percentiles (p10–p90), not fixed score bands. Smoothed: each hex blends parcels around it (Gaussian, widening until it reaches 8 scored parcels), so sharp edges such as a floodway bleed into neighboring hexes. A hex mean can hide a capped parcel; the tooltip counts those inside the hex. Many inputs are tract or block-group values, so neighboring hexes can share them.
Climate & Environment pillar-climate	pillar	value	HouseHack pillar scores (pillars.config.json, default weights) github.com	2026-09-27.1	Distance-weighted mean of nearby City of Pittsburgh parcels, drawn as hexagons (~170 m across, ~85 m when zoomed in), not one parcel	5 caveats 100 = a good place to build new housing. Scores combine observed data using weights that are value judgments. Colors are relative: the bins are City hex percentiles (p10–p90), not fixed score bands. Smoothed: each hex blends parcels around it (Gaussian, widening until it reaches 8 scored parcels), so sharp edges such as a floodway bleed into neighboring hexes. A hex mean can hide a capped parcel; the tooltip counts those inside the hex. Many inputs are tract or block-group values, so neighboring hexes can share them.
Demand pillar-demand	pillar	value	HouseHack pillar scores (pillars.config.json, default weights) github.com	2026-09-27.1	Distance-weighted mean of nearby City of Pittsburgh parcels, drawn as hexagons (~170 m across, ~85 m when zoomed in), not one parcel	5 caveats 100 = a good place to build new housing. Scores combine observed data using weights that are value judgments. Colors are relative: the bins are City hex percentiles (p10–p90), not fixed score bands. Smoothed: each hex blends parcels around it (Gaussian, widening until it reaches 8 scored parcels), so sharp edges such as a floodway bleed into neighboring hexes. A hex mean can hide a capped parcel; the tooltip counts those inside the hex. Many inputs are tract or block-group values, so neighboring hexes can share them.
Overall pillar-overall	pillar	value	HouseHack pillar scores (pillars.config.json, default weights) github.com	2026-09-27.1	Distance-weighted mean of nearby City of Pittsburgh parcels, drawn as hexagons (~170 m across, ~85 m when zoomed in), not one parcel	5 caveats 100 = a good place to build new housing. Scores combine observed data using weights that are value judgments. Colors are relative: the bins are City hex percentiles (p10–p90), not fixed score bands. Smoothed: each hex blends parcels around it (Gaussian, widening until it reaches 8 scored parcels), so sharp edges such as a floodway bleed into neighboring hexes. A hex mean can hide a capped parcel; the tooltip counts those inside the hex. Many inputs are tract or block-group values, so neighboring hexes can share them.
Site Feasibility pillar-site	pillar	value	HouseHack pillar scores (pillars.config.json, default weights) github.com	2026-09-27.1	Distance-weighted mean of nearby City of Pittsburgh parcels, drawn as hexagons (~170 m across, ~85 m when zoomed in), not one parcel	5 caveats 100 = a good place to build new housing. Scores combine observed data using weights that are value judgments. Colors are relative: the bins are City hex percentiles (p10–p90), not fixed score bands. Smoothed: each hex blends parcels around it (Gaussian, widening until it reaches 8 scored parcels), so sharp edges such as a floodway bleed into neighboring hexes. A hex mean can hide a capped parcel; the tooltip counts those inside the hex. Many inputs are tract or block-group values, so neighboring hexes can share them.
Bank branches places-banks	places	observed	FDIC BankFind branch locations api.fdic.gov	FDIC index Sep 2026	Branch locations	2 caveats Credit unions (NCUA) aren't included. Location only; straight-line distance ignores hills, rivers and road network.
Bike share (POGOH) places-bike-share	places	observed	POGOH station information (GBFS feed) pittsburgh.publicbikesystem.net	Snapshot Sep 2026	Station locations	1 caveats Location only; straight-line distance ignores hills, rivers and road network.
Child care places-child-care	places	observed	PA DHS child care providers (data.pa.gov) data.pa.gov	Updated daily; pulled Sep 2026	Provider locations	1 caveats Location only; straight-line distance ignores hills, rivers and road network.
City pools & rec centers places-citiparks	places	observed	Citiparks facilities (City of Pittsburgh ArcGIS) services1.arcgis.com	Pulled Sep 2026	Facility locations, City only	2 caveats Pools are seasonal; hours shown are as published. Location only; straight-line distance ignores hills, rivers and road network.
Colleges & universities places-colleges	places	observed	NCES EDGE postsecondary school locations nces.ed.gov	2023–24	Campus locations	1 caveats Location only; straight-line distance ignores hills, rivers and road network.
Community centers places-community-centers	places	observed	OpenStreetMap contributors (ODbL) merged with Allegheny County Assets (WPRDC) openstreetmap.org	OSM pulled Sep 2026; Assets entries date from 2017–2020	Listed locations	2 caveats Listed, not verified open today. Location only; straight-line distance ignores hills, rivers and road network.
Community health centers places-health-centers	places	observed	HRSA health center service delivery sites data.hrsa.gov	HRSA daily file, Sep 2026	Service site locations	2 caveats Administrative-only sites are excluded; 3 sites are mobile vans. Location only; straight-line distance ignores hills, rivers and road network.
Dentists places-dentists	places	observed	OpenStreetMap contributors (ODbL) merged with Allegheny County Assets (WPRDC) openstreetmap.org	OSM pulled Sep 2026; Assets entries date from 2017–2020	Listed locations	3 caveats Listed, not verified open today. Doesn't say which dentists take Medicaid. Pair with the dental shortage areas layer. Location only; straight-line distance ignores hills, rivers and road network.
Drug & alcohol treatment places-treatment	places	observed	PA Department of Health licensed drug & alcohol facilities (PASDA) mapservices.pasda.psu.edu	Dec 2025 release	Facility locations	1 caveats Location only; straight-line distance ignores hills, rivers and road network.
EMS stations places-ems	places	observed	Allegheny County EMS agencies (ConnectGovs GIS) + City of Pittsburgh EMS stations + OpenStreetMap (ODbL) services.arcgis.com	County list Feb 2024; City layer and OSM Sep 2026	Agency bases, clipped to the county	2 caveats Agency headquarters or base addresses, not every sub-station. Location only; straight-line distance ignores hills, rivers and road network.
Farmers markets places-farmers-markets	places	observed	WPRDC farmers markets (CC0) data.wprdc.org	Updated Aug 2026	Market locations	2 caveats Most are seasonal. Location only; straight-line distance ignores hills, rivers and road network.
Fire stations places-fire	places	observed	Allegheny County fire departments (ConnectGovs GIS), City of Pittsburgh fire stations, and OpenStreetMap (ODbL) services.arcgis.com	County list Jan 2024; City layer and OSM Sep 2026	Station locations, clipped to the county	2 caveats Sources merged within 200 m; union halls, museums and training sites excluded. Location only; straight-line distance ignores hills, rivers and road network.
Food pantries & distributions places-food-banks	places	observed	Greater Pittsburgh Community Food Bank partner programs (hosted by Pittsburgh Regional Transit) services3.arcgis.com	Data gathered Apr 2025	Program locations	2 caveats Hours and eligibility change often; check the Food Bank's locator (pittsburghfoodbank.org) before going. Location only; straight-line distance ignores hills, rivers and road network.
Grocery stores places-groceries	places	observed	USDA FNS SNAP authorized retailers merged with ACHD food permits (deduplicated) services1.arcgis.com	SNAP list Sep 2026; ACHD permits as of 2025	Store locations	4 caveats Store tier comes from USDA and county permit categories, not a store audit; dollar stores, gas stations and beer distributors are never counted as grocery stores. County-permit-only stores are kept only if inspected since 2023; SNAP-listed stores are kept regardless. 'Not matched to a SNAP-authorized store' can mean the name or location didn't match the USDA list, not that the store refuses SNAP. Location only; straight-line distance ignores hills, rivers and road network.
Hospitals places-hospitals	places	observed	PA Department of Health licensed hospitals (PASDA) mapservices.pasda.psu.edu	DOH Hospitals, Nov 2025	Facility locations, Allegheny County	4 caveats Specialty is inferred from the name. Emergency department and star rating come from CMS Care Compare (Jul 2026), matched by address; specialty hospitals are not rated. The CMS star rating is a composite of reported measures, not a full quality judgment. Location only; straight-line distance ignores hills, rivers and road network.
Laundromats places-laundromats	places	observed	OpenStreetMap contributors (ODbL) merged with Allegheny County Assets (WPRDC) openstreetmap.org	OSM pulled Sep 2026; Assets entries date from 2017–2020	Listed locations	3 caveats Listed, not verified open today. Some County Assets entries (2017–20) may have closed. Location only; straight-line distance ignores hills, rivers and road network.
Libraries places-libraries	places	observed	Allegheny County libraries (Allegheny County GIS) services1.arcgis.com	Edited Dec 2024	Library locations	1 caveats Location only; straight-line distance ignores hills, rivers and road network.
Low food access (USDA) food-access	places	observed	USDA ERS Food Access Research Atlas 2025 ers.usda.gov	2025 edition (store and ACS inputs roughly 2019–2021)	Census tract (2020)	2 caveats Straight-line distance to a supermarket, not walking distance; hills and rivers make real trips longer. Only flagged tracts are drawn. Pair with the grocery store points.
Magisterial district courts places-courts	places	observed	Allegheny County Magisterial District Courts (ConnectGovs GIS) services.arcgis.com	Edited Dec 2023	Court office locations	2 caveats Pairs with the eviction filings layer. Location only; straight-line distance ignores hills, rivers and road network.
Nursing homes places-nursing-homes	places	observed	PA Department of Health nursing homes (PASDA) mapservices.pasda.psu.edu	Sep 2026 release	Facility locations	2 caveats The Legal feasibility senior-housing layer also shows the City's nursing homes. Location only; straight-line distance ignores hills, rivers and road network.
Other food retail places-food-other	places	observed	USDA FNS SNAP authorized retailers merged with ACHD food permits (deduplicated) services1.arcgis.com	SNAP list Sep 2026; ACHD permits as of 2025	Store locations	2 caveats Tier comes from USDA and county permit categories, not a store audit. Location only; straight-line distance ignores hills, rivers and road network.
Parks & greenways parks	places	observed	Allegheny County GIS: municipal parks, county park boundaries, greenways services1.arcgis.com	Municipal parks and greenways edited Apr 2026; county parks Jun 2023	Park polygons; measure distance to the nearest edge, not the center	2 caveats Park quality and entrances vary. Straight-line distance ignores hills and rivers.
Pharmacies places-pharmacies	places	observed	CMS NPPES NPI Registry (pharmacy organizations) npiregistry.cms.hhs.gov	Registry pulled Sep 2026	Pharmacy locations (geocoded addresses)	3 caveats Registry listings: NPPES doesn't record closures, so a few listed pharmacies may be gone. Rite Aid (84 registrations in Allegheny ZIP codes) is excluded: the chain closed all stores by Sep 2025. Location only; straight-line distance ignores hills, rivers and road network.
Playgrounds (City) places-playgrounds	places	observed	City of Pittsburgh playgrounds (WPRDC, CC-BY) data.wprdc.org	Pulled Sep 2026	Playground locations, City only	1 caveats Location only; straight-line distance ignores hills, rivers and road network.
Police stations places-police	places	observed	Allegheny County municipal police (ConnectGovs GIS) + OpenStreetMap (ODbL) + PA State Police stations (PASDA) services.arcgis.com	County list Jan 2024; OSM Sep 2026; PSP Mar 2025	Station locations, clipped to the county	3 caveats One point per department headquarters; some departments cover several municipalities (listed under 'Serves'). Pittsburgh zone stations come from OSM. Location only; straight-line distance ignores hills, rivers and road network.
Post offices places-post-offices	places	observed	OpenStreetMap contributors (ODbL) merged with Allegheny County Assets (WPRDC) openstreetmap.org	OSM pulled Sep 2026; Assets entries date from 2017–2020	Listed locations	2 caveats Listed, not verified open today. Location only; straight-line distance ignores hills, rivers and road network.
Regional trails trails	places	observed	Allegheny County regional trails (county planning) services1.arcgis.com	Edited Jan 2025	Trail segments	1 caveats Planned and proposed segments are not built.
Restaurants & shops (density) commerce-density	places	observed	Allegheny County Health Dept food facility permits (WPRDC, CC0) data.wprdc.org	Permits as of 2025	Density of ~11k licensed restaurants and convenience/retail food stores	2 caveats Food-licensed businesses only; other shops are not counted. Closures can lag.
School performance (PA Future Ready) school-quality	places	observed	PA Department of Education, Future Ready PA Index 2024–25 futurereadypa.org	2024–25 school year	School locations (cyber schools excluded)	4 caveats Proficiency closely tracks family income (r ≈ −0.87 across Allegheny schools); growth reflects what the school adds. Schools are not color-ranked by scores on purpose. PVAAS growth score is on a 50–100 scale (higher = more growth than the statewide average for similar students); there is no pass/fail line. Attendance zones differ from enrollment (charters, magnets, district lines).
Schools places-schools	places	observed	Allegheny County Schools (Allegheny County GIS) services1.arcgis.com	Edited Mar 2026	School locations	2 caveats Locations only; schools are deliberately not ranked by test scores. Location only; straight-line distance ignores hills, rivers and road network.
Senior centers places-senior-centers	places	observed	Allegheny County Assets (WPRDC, CC0) data.wprdc.org	Registry updated Sep 2026; many underlying sources date from 2017–2020	Listed locations	2 caveats Listed, not verified open today. Location only; straight-line distance ignores hills, rivers and road network.
Subsidized housing subsidized-housing	places	observed	HUD: LIHTC database, Public Housing Buildings, Multifamily Assisted Properties services.arcgis.com	LIHTC Dec 2024; public housing and assisted multifamily Jul 2026	Property / building locations, clipped to the county	3 caveats A property can appear in more than one program (e.g. LIHTC + Section 8), so don't sum units across programs. The LIHTC 'may be ending' flag assumes a 30-year affordability period from the placed-in-service year. HUD geocodes can be approximate.
Urgent care & other clinics places-clinics	places	observed	CMS NPPES NPI Registry (clinic taxonomies) npiregistry.cms.hhs.gov	Registry pulled Sep 2026	Clinic locations (geocoded addresses)	3 caveats Health-center clinics within ~100 m of an HRSA site are dropped as duplicates. NPPES doesn't record closures. Location only; straight-line distance ignores hills, rivers and road network.
1937 redlining map (HOLC) holc-1937	policy	observed	Mapping Inequality, Nelson et al., University of Richmond Digital Scholarship Lab (CC BY-SA), via WPRDC dsl.richmond.edu	1937	Historic appraisal areas (1937 city and some inner suburbs)	2 caveats Historic and racially discriminatory grades, shown for context on how today's patterns formed. Not a description of these places today.
City zoning overlays & historic districts city-zoning-overlays	policy	policy	City of Pittsburgh zoning overlays and historic districts (City ArcGIS) services1.arcgis.com	Pulled Sep 2026	Overlay districts	2 caveats City of Pittsburgh only. Pending zoning bills (e.g. 2025-1545, 2026-0834) may change these; check current code.
Community organizations (RCOs) rcos	policy	observed	City of Pittsburgh Registered Community Organizations (City ArcGIS) services1.arcgis.com	Edited Aug 2026	RCO service areas (they can overlap)	1 caveats Contact names, emails and phone numbers are intentionally not shown; use the organization's website.
Health professional shortage areas health-shortage-areas	policy	policy	HRSA Health Professional Shortage Areas (federal designations) data.hrsa.gov	Updated daily; pulled Sep 2026	Designated service areas that touch Allegheny County	3 caveats Faint fill = proposed for withdrawal. Where an area has a current designation and an older record proposed for withdrawal, only the current one is shown. No mental-health shortage areas are currently designated here.
Subsidy designation areas designation-areas	policy	policy	HUD: Qualified Census Tracts 2026, Difficult Development Areas 2026, Opportunity Zones huduser.gov	QCT/DDA 2026 (edited Sep 2025); Opportunity Zones designated 2018	Tracts (QCT, OZ) and ZIP areas (DDA)	2 caveats LIHTC projects in a QCT or DDA can receive up to a 30% eligible-basis boost. Opportunity Zones use 2010 tracts; newer designations are not reflected.

§09 · datasets

Dataset catalog

All 134 datasets and services behind the map and the scores: publisher, exact endpoint, vintage, license and where each one is used. Compiled from the code on every branch and checked so that every endpoint the code calls is cited. Accessed 2026-09-26 unless noted.

134 of 134

Dataset (publisher)	Source / endpoint	Vintage	License	Used in
Allegheny County parcel boundaries (Allegheny County GIS, via WPRDC)	data.wprdc.org (~585k parcels; imported into the team's Neon PostGIS database, reprojected from EPSG:2272)	Sept 2026	WPRDC (county data)	Parcel outlines on the map; parcel selection
City of Pittsburgh `ParcelsPublic` (City of Pittsburgh)	`CITY/ParcelsPublic/FeatureServer/0` (142,635 features, 142,365 distinct PINs; private owners already anonymized as "Private")	edited 2026-09-21	City open data	Pillar "parcel spine" (every scored parcel), vacant-building address join, lot area
Parcel centroids with census geographies (Allegheny County, via WPRDC)	data.wprdc.org (585,387 rows; 2020 tract and block-group codes, verified 99.994% match)	March 2025	CC0	Parcel → tract / block-group crosswalk for every pillar indicator; parcel-level heat and canopy sampling
City zoning districts `PGHWebZoning` (City of Pittsburgh Dept. of City Planning)	`CITY/PGHWebZoning/FeatureServer/0` (1,069 polygons, field `zon_new`); also WPRDC data.wprdc.org	edited 2026-09-22	City open data	`pittsburgh-zoning.geojson` base layer, `residential-zoning`, legal matrix join, permit and ZBA district joins
City neighborhoods `PGHWebNeighborhoods`	`CITY/PGHWebNeighborhoods/FeatureServer/0` (90 polygons)	edited 2026-07-02	City open data	Permit and ZBA neighborhood joins
City boundary	`CITY/City_Boundary/FeatureServer/0`	–	City open data	Clipping City-only layers
Allegheny County boundary	`COUNTY/Allegheny_County_Boundary/FeatureServer/0`; TIGERweb tigerweb.geo.census.gov (GEOID 42003)	current	County / public domain	Clipping every county-wide layer
Census tracts, block groups (2020) and ZCTAs (U.S. Census Bureau)	Census Reporter geometry API api.censusreporter.org (1,062 block groups) and `140\|05000US42003` (394 tracts); TIGERweb ZCTA tigerweb.geo.census.gov	2020 geography	Public domain	Every tract / block-group / ZIP choropleth
Census tracts (2010 geography)	tigerweb.geo.census.gov	2010	Public domain	Opportunity Atlas, life expectancy and blood-lead layers (2010-tract data)
2020↔2010 tract relationship file; 2010 Gazetteer tract file (U.S. Census Bureau)	www2.census.gov ; www2.census.gov	2020 / 2010	Public domain	Early population-change estimate (superseded by §2.6 household growth)
Basemap: CARTO Dark Matter and Positron vector styles, incl. 3D building extrusions	basemaps.cartocdn.com ; basemaps.cartocdn.com (buildings from the `building` source-layer, OpenStreetMap data)	live	CARTO terms + ODbL	App basemap; 3D buildings (Mat's `3d-buildings` branch)
Basemap: OpenStreetMap raster tiles	tile.openstreetmap.org	live	ODbL; OSM tile usage policy	Alternate basemap
U.S. Census Geocoder	geocoding.geo.census.gov and `/geocoder/geographies/onelineaddress` (benchmark Public_AR_Current)	live	Public domain	Geocoding pharmacies, clinics, senior/care facilities
OpenStreetMap Nominatim	nominatim.openstreetmap.org	live	ODbL; Nominatim usage policy	App address search; fallback geocoding for 29 pharmacies
National Flood Hazard Layer, flood hazard zones (FEMA)	hazards.fema.gov (county bbox; 9,501 polygons kept after dropping "minimal hazard")	effective FIRMs, current	Public domain	`flood-zones`; pillar gates (floodway, floodplain)
National Risk Index, census tracts (FEMA)	services.arcgis.com (inland flood, heat wave, cold wave, winter weather, strong wind, landslide, ice storm, tornado ratings)	NRI Dec 2025	Public domain	`weather-risk`; climate pillar
3D Elevation Program slope, computed on the fly (USGS)	elevation.nationalmap.gov (raster function `Slope Degrees`, remapped to 15–25% and ≥25%)	service dated 2026-08-25	Public domain	`slope-raster`
Steep slopes ≥25% `PGHWebSlope25` (City)	`CITY/PGHWebSlope25/FeatureServer/0` (1,714 polygons)	edited 2018-05-31	City open data	`city-steep-slopes`; pillar gate (steep slope share)
Landslide-prone areas `PGHWebLandslideProne` (City)	`CITY/PGHWebLandslideProne/FeatureServer/0` (37 polygons)	edited 2024-12-19	City open data	`city-hazard-overlays`; pillar gate
Undermined areas `PGHWebUndermined` (City)	`CITY/PGHWebUndermined/FeatureServer/0` (47 polygons)	edited 2018-04-27	City open data	`city-hazard-overlays`; pillar flag
Landslide susceptibility (Allegheny County GIS; flags consistent with 1970s–80s USGS Pomeroy mapping, provenance inferred)	`COUNTY/Landslide_Prone_Areas/FeatureServer/0` (7,242 polygons; 6,792 with a hazard flag)	edited 2018-07-19	Not stated	`landslide-susceptibility`; pillar gate
Landslide incidents on county roads (Allegheny County Public Works)	`COUNTY/Landslides_DPW/FeatureServer/0` (115 points)	edited 2025-09-24	Not stated	`landslide-incidents`
Landslide public-assistance sites (Allegheny County)	`COUNTY/Landslide_Public_Assistance/FeatureServer/0` (86 points)	edited 2023-10-20	Not stated	`landslide-public-assistance`
Mined-out areas, underground coal (PA DEP)	gis.dep.pa.gov (997 polygons in the county bbox)	current	PA public data	`mined-out-areas`
Mine Subsidence Insurance risk, mining confirmed (PA DEP)	gis.dep.pa.gov (raster export)	current	PA public data	`mine-subsidence-raster`
Tornado paths 1950–2025 (NOAA Storm Prediction Center)	spc.noaa.gov (44 tracks touching the county)	through 2025	Public domain	`tornado-paths`
Urban heat island, 2020 mean land surface temperature tiles (City of Pittsburgh)	tiles.arcgis.com (display only; no values)	2020	City open data	`surface-heat-raster`
Landsat 8 Collection 2 Level-2 surface temperature (USGS; served by Microsoft Planetary Computer)	STAC planetarycomputer.microsoft.com collection `landsat-c2-l2`; data API `/api/data/v1/item/bbox/...npy` (band `lwir11` + `qa_pixel`). Scenes: LC08_L2SP_017032_20250810_02_T1, LC08_L2SP_017032_20240823_02_T1, LC08_L2SP_017032_20230618_02_T1, LC08_L2SP_017032_20230602_02_T1	2023–2025 summers	Public domain (USGS Landsat)	Pillar indicators: parcel and block-group land surface temperature (4-scene anomaly composite)
NLCD 2021 tree canopy cover (USFS / MRLC)	mrlc.gov ; WMS mrlc.gov (layer `nlcd_tcc_conus_2021_v2021-4`); WCS mrlc.gov and `/geoserver/mrlc_display/wcs` (coverage `mrlc_display__nlcd_tcc_conus_2021_v2021-4`)	2021	Public domain	`tree-canopy-raster`; pillar canopy indicators
NLCD 2021 percent developed impervious (MRLC)	WMS layer `NLCD_2021_Impervious_L48`; WCS coverage `mrlc_download__NLCD_2021_Impervious_L48`	2021	Public domain	`impervious-raster`; pillar impervious indicator
EPA EJScreen v2.32, state percentiles by block group (EPA; via the Public Environmental Data Partners mirror after EPA took EJScreen offline in 2025)	services2.arcgis.com (PM2.5, ozone, NO2, diesel PM, traffic proximity, RSEI air toxics)	v2.32 (indicator years vary)	Public domain (EPA data; mirror)	`air-quality`; climate pillar air indicators. Mirror homes: pedp-ejscreen.azurewebsites.net , screening-tools.com ; the original ejscreen.epa.gov was offline
ACHD point-source emissions inventory (Allegheny County Health Dept, via WPRDC)	data.wprdc.org (resource `1ab77bb5-5684-430e-bdda-fb401167fb6a`, 16,768 rows, 58 current facilities)	2010–2021 (latest year 2021)	WPRDC	Pillar indicator: nearby industrial emissions (`achd_facilities_latest.csv`)
Water service line material, lead (Pittsburgh Water / PWSA)	`PWSA/PGH2O_Water_Service_Line_Material/FeatureServer/0` (80,876 service addresses)	edited 2026-08-31	Not stated; PWSA accuracy disclaimer	`lead-service-lines`; pillar flag
Sewer lines, county-wide (Allegheny County GIS; 3 Rivers Wet Weather "LBs" export)	`COUNTY/Pittsburgh_Sewers/FeatureServer/0` (125,128 segments, 86 municipal systems)	2012 export	Not stated	`sewer-lines` (loaded by viewport)
Pittsburgh Regional Transit stops with scheduled trips (PRT, via WPRDC)	data.wprdc.org ; file data.wprdc.org (6,389 stops)	GTFS feed 2606, updated 2026-09-22	CC BY	`transit-stops`; access pillar transit indicator
Access Across America: Transit 2024, jobs reachable by transit (University of Minnesota Accessibility Observatory)	conservancy.umn.edu (PA file, block groups, 30 / 45 minutes)	2024	CC BY-NC 4.0	`jobs` (transit_jobs_30/45); access pillar
POGOH bike-share stations (POGOH, GBFS feed)	pittsburgh.publicbikesystem.net (60 stations)	live	GBFS public feed	`places-bike-share`
PennDOT reportable crashes, Allegheny County (PennDOT, via WPRDC)	data.wprdc.org (2025 resource `c6bedb17-2d23-49b1-843a-8f6b41e5e5c3`; 2024 `4c016b4c-59f0-45ca-981c-718c784b3462`; 2023 `96777349-57df-48fb-a1d7-8384786fe71a`)	2023–2025	CC0	`serious-crashes` (killed-or-seriously-injured, 1,174)
American Community Survey 5-year (U.S. Census Bureau, via the Census Reporter API)	api.censusreporter.org (human-readable: censusreporter.org ; table pages such as censusreporter.org) — tables B19013 (median household income), B25064 (median gross rent), B25077 (home value), B25070 (rent burden), B25002/B25004 (vacancy), B25003 (tenure), B25001 (housing units), B28002 (internet), B01001/B01003 (age, population), B25010 (household size), B11001 (household type), B14007 (school enrollment), B26001 (group quarters)	ACS 2020–2024	Public domain	`housing-costs`, `home-internet`; pillar demand, need and affordability indicators; college-enrollment adjustment
Comprehensive Housing Affordability Strategy, CHAS (HUD)	huduser.gov (tract Table 8) + data dictionary huduser.gov ; landing page huduser.gov	2018–2022 (released Dec 2025)	Public domain	`chas-cost-burden`; need pillar
FY2026 income limits (HUD)	huduser.gov (Pittsburgh HMFA, fips 4200399999; MFI $110,400); also the HUD USER API huduser.gov (needs a HUD USER token; the xlsx was used)	FY2026	Public domain	`hud_fy2026_pittsburgh.json` (affordability context)
FY2026 Fair Market Rents and Small Area FMRs (HUD)	huduser.gov ; huduser.gov (393 ZIPs)	FY2026	Public domain	`pittsburgh_safmr_fy2026.csv`
Designated mandatory Small Area FMR areas (HUD)	huduser.gov (Pittsburgh HMFA designated 2016, implemented 4/1/2018)	Aug 2024 list	Public domain	Methodology note (vouchers use ZIP-level SAFMRs)
FY2026 Multifamily Tax Subsidy Project (LIHTC) income limits (HUD)	huduser.gov ; huduser.gov	FY2026	Public domain	`hud_fy2026_pittsburgh.json`
Location Affordability Index v3 (HUD / DOT)	`HUD/Location_Affordability_Index_v3/FeatureServer/0` + methodology files.hudexchange.info (Table 11 household profiles); hudexchange.info ; hudgis-hud.opendata.arcgis.com	2012–2016 ACS inputs; 2010 tracts	Public domain	`location-affordability` (VMT, housing + transportation cost); pillar indicators
Qualified Census Tracts 2026, Difficult Development Areas 2026, Opportunity Zones (HUD)	`HUD/QUALIFIED_CENSUS_TRACTS_2026/FeatureServer/0`; `HUD/Difficult_Development_Areas_2026/FeatureServer/0`; `HUD/Opportunity_Zones/FeatureServer/13`; landing huduser.gov	2026 (OZ 2018 designations)	Public domain	`designation-areas`
LIHTC properties (HUD)	`HUD/LIHTC/FeatureServer/0`; huduser.gov	data edited 2024-12-09	Public domain	`subsidized-housing`; senior housing
Public Housing Buildings and Developments (HUD)	`HUD/Public_Housing_Buildings/FeatureServer/0`; `HUD/Public_Housing_Developments/FeatureServer/0`	edited 2026-07-21	Public domain	`subsidized-housing`; senior housing (HACP units)
Multifamily Properties – Assisted; Section 202; Section 811 (HUD)	`HUD/MULTIFAMILY_PROPERTIES_ASSISTED/FeatureServer/0`; `HUD/HUD_Section_202_Properties/FeatureServer/0`; `HUD/Section_811_Properties/FeatureServer/21`	edited 2026-07-21	Public domain	`subsidized-housing` (incl. HAP contract expirations); `senior-supportive-housing`
Housing Choice Vouchers by tract (HUD)	`HUD/Housing_Choice_Vouchers_by_Tract/FeatureServer/0`	edited 2026-07-21	Public domain	`housing-vouchers`
Housing Authority of the City of Pittsburgh communities (HACP)	hacp.org (16 senior communities)	read 2026-09-26	Web page	`senior-supportive-housing`
Low-Income Energy Affordability Data (LEAD) Tool, 2022 (U.S. DOE)	data.openei.org (file "PA AMI Census Tracts 2022.csv"); DOI 10.25984/2504170; units per Scheier & Kittner, Nature Communications 2022	ACS 2022 basis	CC BY 4.0	Pillar energy-burden indicator
Eviction Tracking System, Pittsburgh (Princeton Eviction Lab; filings collected by Legal Services Corporation with CMU CREATE Lab)	eviction-lab-data-downloads.s3.amazonaws.com (Pittsburgh rows run 01/2020–12/2025, ZIP level); pages evictionlab.org , evictionlab.org	2025 filings	Eviction Lab terms	`evictions` (ZIP); pillar displacement indicator
Market Value Analysis 2021 and Displacement Risk Ratio 2021 (Reinvestment Fund for URA / Allegheny County, via WPRDC)	data.wprdc.org (package `f669d677-c9e2-4d2f-b16f-c9ab5a4f3d10`) — MVA GeoJSON `.../resource/ec09f5ad-f43e-4f06-af3a-65641c2818dc/download/mva.geojson`; DRR shapefile `.../resource/3a648531-15d7-4fc0-aba0-611968b92435/download/pitts_allegheny_drr2021.zip`; dictionary `.../resource/45fe19b5-26bd-455e-aae7-c4b99f30946d/...`	2017–2019 sales; 2014–2020 DRR	CC0	`market-mva`; demand and displacement indicators
Property sale transactions (Allegheny County, via WPRDC)	Resource `5bbe6c55-bce6-4edb-9d04-68edeb6bf7b1`, data.wprdc.org (503,747 rows; filtered to `VALID SALE`, dwellings, ≥ $10,000)	2012–Sept 2026	CC0	`market-zip` (median sale price); `allegheny_tract_sales_2019_2025.csv`; demand pillar turnover
Zillow Observed Rent Index, ZIP (Zillow Research)	zillow.com ; ZORI files.zillowstatic.com (55 Allegheny ZIPs, Aug 2026); ZHVI files.zillowstatic.com	monthly through 2026-08-31	Zillow Research terms	`market-zip`; demand pillar rent growth
USPS vacant addresses by tract (HUD/USPS, via WPRDC)	data.wprdc.org (2023-Q4 used)	2023	CC0	`usps-vacancy`; demand pillar
Property assessments (Allegheny County Office of Property Assessments, via WPRDC)	data.wprdc.org (API resource `65855e14-549e-4992-b5be-d629afc676fa`; bulk CSV resource `9a1c60bd-f9f7-4aba-aeb7-af8c3aaa44e5`)	as of 2026-09-01	CC0	`vacant-lots` (75,982 vacant-land parcels); dwelling filter for tract sales
USPS-vacant parcels `Vacant_USPS_Feb_24` (City of Pittsburgh)	`CITY/Vacant_USPS_Feb_24/FeatureServer/0` (4,686 parcels; 4,677 on the map), joined to `ParcelsPublic` addresses	Feb 2024 snapshot; edited 2026-05-28	City open data	`vacant-buildings`
City-owned property (City of Pittsburgh, via WPRDC)	data.wprdc.org (resource `e1dcee82-9179-4306-8167-5891915b62a7`, 12,477 parcels; 12,454 on the map)	updated daily (2026-09-26)	CC BY	`city-owned-land`
City and school property tax delinquency (City of Pittsburgh, via WPRDC)	data.wprdc.org (resource `ed0d1550-c300-4114-865c-82dc7c23235b`)	2026-09-24	CC BY	`tax-delinquent`
Treasurer sales (City of Pittsburgh, via WPRDC)	data.wprdc.org (resource `6b2aa631-26e0-4d02-abe0-7fb87707210c`, 96 parcels)	2026-09-24	Not specified	`treasury-sales`
Condemned and dead-end properties (City of Pittsburgh PLI, via WPRDC)	data.wprdc.org (resource `0a963f26-eb4b-4325-bbbc-3ddf6a871410`, 3,569)	daily	CC BY	`condemned-properties`
PLI / DOMI / ES code violations (City of Pittsburgh, via WPRDC)	data.wprdc.org (resource `70c06278-92c5-4040-ab28-17671866f81c`; last 24 months aggregated per block group)	daily	CC0	`code-violations`
County real estate tax delinquency; mortgage foreclosure filings (Allegheny County, via WPRDC)	data.wprdc.org (resources `b81a8a22-…` 2025, `96e9d6b2-…` cumulative); data.wprdc.org (`859bccfd-0e12-4161-a348-313d734f25fd`)	2026-09	CC0	Researched and handed off; not on the map (need a parcel-centroid join)
2020 Census redistricting data, PL 94-171, Pennsylvania (U.S. Census Bureau)	www2.census.gov (H1 occupied units, P1 population; tract level)	April 2020	Public domain	Pillar household-growth indicator (`allegheny_tract_household_growth_2020_2024.csv`)
LEHD Origin-Destination Employment Statistics, LODES8 (U.S. Census Bureau)	WAC lehd.ces.census.gov and `..._2019.csv.gz`; RAC `.../pa/rac/pa_rac_S000_JT00_2023.csv.gz`; OD `.../pa/od/pa_od_main_JT00_2023.csv.gz` and `pa_od_aux_JT00_2023.csv.gz`; crosswalk `.../pa/pa_xwalk.csv.gz`	2019 and 2023	Public domain	`jobs`; demand pillar job growth (`allegheny_jobs_demand_bg_2023.csv`)
Quarterly Census of Employment and Wages, Allegheny County (BLS)	data.bls.gov (and 2026/1)	2025 Q4	Public domain	Context statistic (county jobs and wages)
New residential construction permits (City of Pittsburgh PLI, via WPRDC)	data.wprdc.org (resource `f4d1177a-f597-4c32-8cbf-7885f56253f6`, 65,378 permits, 2019-06 → 2026-09)	daily	CC BY	`permits-activity`; `pgh_new_residential_permits_classified.csv` (keyword-classified housing type; demand-pillar new-construction indicator, labeled an assumption); raw rows in `pgh_new_construction_permits_raw.csv`
Social Vulnerability Index 2022 (CDC / ATSDR)	services3.arcgis.com (394 tracts); atsdr.cdc.gov	2022	Public domain	`equity`
Child Opportunity Index 3.0 (diversitydatakids.org, via a third-party ArcGIS mirror)	services7.arcgis.com (392 tracts)	2021 data year	diversitydatakids terms	`equity`
The Opportunity Atlas (Opportunity Insights / U.S. Census Bureau)	opportunityinsights.org ; 2010→2020 tract crosswalk opportunityinsights.org ; opportunityinsights.org	1978–83 birth cohorts; 2010 tracts	Public	`opportunity-atlas`
HOLC 1937 residential security map, Pittsburgh (Mapping Inequality, University of Richmond; via WPRDC)	data.wprdc.org (full-detail version: resource `e17b6d76-4060-4691-be66-2edc367d87b8`, papittsburgh1937.geojson); dsl.richmond.edu	1937	CC BY-SA	`holc-1937`
Food Access Research Atlas 2025 (USDA ERS)	gisportal.ers.usda.gov (394 tracts; 32 low-income / low-access); ers.usda.gov	2025 edition	Public domain	`food-access`
National Walkability Index (U.S. EPA, Smart Location Database)	geodata.epa.gov (1,100 block groups); epa.gov	~2019 inputs	Public domain	`walkability`
Census college enrollment and group quarters (ACS B14007, B26001)	Census Reporter, as in §2.4	ACS 2020–2024	Public domain	Pillar student-household adjustment (`allegheny_tract_college_enrollment.csv`)
PLACES: census tract estimates, 2024 release (CDC)	data.cdc.gov (382 tracts); cdc.gov	2024 release	Public domain	`health-outcomes`
U.S. Small-Area Life Expectancy Estimates, USALEEP (CDC / NCHS)	ftp.cdc.gov (357 Allegheny tracts); cdc.gov	2010–2015	Public domain	`life-expectancy`
Health Professional Shortage Areas: primary care, dental, mental health (HRSA)	gisportal.hrsa.gov (layers 2, 6, 10, 11); data.hrsa.gov ; bulk file data.hrsa.gov	updated daily	Public domain	`health-shortage-areas` (7 areas)
Health center service delivery and look-alike sites (HRSA)	data.hrsa.gov (35 active Allegheny sites); download page data.hrsa.gov	2026-09-26	Public domain	`places-health-centers`; access pillar
Hospital General Information, star ratings and ER flag (CMS Care Compare)	data.cms.gov (18 Allegheny hospitals)	dataset modified 2026-07-22	Public domain	`places-hospitals` (CMS rating, emergency flag)
Licensed hospitals (PA Department of Health, via PASDA)	`PASDA_DOH/6` ("DOH Hospitals 202511", 26 in Allegheny)	Nov 2025	PA public data	`places-hospitals`
Licensed drug and alcohol treatment facilities (PA DOH, via PASDA)	`PASDA_DOH/3` ("202512", 70 in Allegheny)	Dec 2025	PA public data	`places-treatment`
Nursing homes (PA DOH, via PASDA and data.pa.gov)	`PASDA_DOH/12` ("202609", 52 in Allegheny); data.pa.gov	Sept 2026	PA public data	`places-nursing-homes`; `senior-supportive-housing`
Nursing home provider information, certified beds (CMS)	data.cms.gov (state = PA)	modified 2026-08-01	Public domain	`senior-supportive-housing`
NPPES NPI Registry: pharmacies and clinics (CMS)	npiregistry.cms.hhs.gov (organization NPIs; Pharmacy taxonomies; Urgent Care, FQHC, Primary Care and Community Health clinic taxonomies; 112 Allegheny ZIPs)	live, 2026-09-26	Public domain	`places-pharmacies` (275; Rite Aid removed), `places-clinics` (69); access pillar
Elevated blood lead level rates (Allegheny County Health Dept, via WPRDC)	data.wprdc.org (tract resource `8432f1ad-c5bf-447f-b34b-160d8ee063b6`; 2015–20 column only; neighborhood resource `4e215bd5-7e83-4cd2-ae48-4de5dc29910c`; Lead Data Guide `a8f69277-6328-4421-a508-8b2100d753e3`); ACHD testing page alleghenycounty.us	2015–2020	CC0	`child-blood-lead`
Police incident blotter, UCR-coded archive (Pittsburgh Bureau of Police, via WPRDC)	data.wprdc.org (resource `044f2016-1dfd-4ab0-bc1e-065da05fca2e`, 340,996 incidents; last 3 years to 2023-11)	ends Nov 2023	CC BY	`safety` (violent / property per 1,000, City tracts)
Homicide incidents 2007–2022 and 2024 (Allegheny County, Medical Examiner locations)	`COUNTY/AC_Homicide_Incidents_2007_2022/FeatureServer/2`; `COUNTY/2024_Homicides_in_Allegheny_County/FeatureServer/59` (victim fields dropped; aggregated per tract)	2007–2022, 2024	Not stated	`safety` (homicides per 10k)
PennDOT crashes	See §2.3	2023–2025	CC0	`serious-crashes`
SNAP-authorized retailers (USDA Food and Nutrition Service)	services1.arcgis.com (839 in Allegheny)	edited 2026-09-17	Public domain	`places-groceries` (329), `places-food-other` (1,257); access pillar
Geocoded food facilities and inspection history (Allegheny County Health Dept, via WPRDC)	Facilities resource `112a3821-334d-4f3f-ab40-4de1220b1a0a` (data.wprdc.org); inspections 2014–2025 resource `4ea730d8-2bf9-4783-b0a5-b41ed687097e`; package data.wprdc.org	as of 2025 (inspections to 2025-07-31)	CC0	Grocery merge (stale-record filter); `commerce-density`
Allegheny County Assets (WPRDC aggregator)	data.wprdc.org (resource `5c7825d2-6814-40c7-aefe-3d0f3d6f22e7`, 32,001 assets; sources mostly 2017–2020)	modified 2026-09-21	CC0	`places-senior-centers`; merged into post offices, laundromats, dentists, community centers
OpenStreetMap amenities via the Overpass API	overpass-api.de and overpass.kumi.systems (4 tiles, clipped to the county; 2,329 features)	OSM data as of 2026-05-31	ODbL	Post offices, laundromats, dentists, community centers, police and EMS additions, fire-station cross-check
Bank branches (FDIC BankFind)	api.fdic.gov (364 Allegheny branches)	index built 2026-09-25	Public domain	`places-banks`
Food assistance programs (Greater Pittsburgh Community Food Bank; hosted by PRT)	services3.arcgis.com (163 sites; 162 on the map)	April 2025	Not stated	`places-food-banks`; access pillar
Farmers markets (Allegheny County, via WPRDC)	data.wprdc.org (resource `0d99978a-ccf6-4315-af5a-a95acec87a9f`, 69)	2026-08-01	CC0	`places-farmers-markets`
Child care providers (PA Department of Human Services, data.pa.gov)	data.pa.gov (718 providers)	updated daily	Public	`places-child-care`; access pillar
Allegheny County Schools (Allegheny County GIS)	`COUNTY/Allegheny_County_Schools/FeatureServer/0` (411)	edited 2026-03-27	Not stated	`places-schools`; access pillar
Future Ready PA Index 2024–25 (PA Department of Education)	School Fast Facts futurereadypa.org ; performance futurereadypa.org ; glossary futurereadypa.org	2024–25	Public	`school-quality` (275 schools)
Postsecondary school locations (NCES EDGE)	nces.ed.gov (37); nces.ed.gov	2023–24	Public domain	`places-colleges`
Libraries (Allegheny County GIS)	`COUNTY/Libraries/FeatureServer/0` (72)	edited 2024-12-04	Not stated	`places-libraries`; access pillar
Municipal police departments; designated EMS agencies; fire departments; magisterial district court offices (ConnectGovs)	`CG/Police_Departments_Allegheny_County/FeatureServer/0`; `CG/EMS_Departments_Allegheny_County/FeatureServer/0`; `CG/Fire_Departments_Allegheny_County/FeatureServer/0`; `CG/Magisterial_District_Judges_Office_Locations/FeatureServer/0`	2023-12 to 2024-02	Not stated	`places-police` (122), `places-ems` (58), `places-fire` (252), `places-courts` (46)
City fire stations and EMS medic stations (City of Pittsburgh)	`CITY/Fire_Station/FeatureServer/0`; `CITY/EMS_Station/FeatureServer/0`	2024-05	City open data	`places-fire`, `places-ems`
Pennsylvania State Police stations (PSP, via PASDA)	mapservices.pasda.psu.edu	March 2025	PA public data	`places-police`
Parks, county parks, greenways (Allegheny County GIS)	`COUNTY/Allegheny_County_Municipal_Parks_New/FeatureServer/0` (752); `COUNTY/ParkBoundaries/FeatureServer/0` (10); `COUNTY/Allegheny_County_Greenways_New/FeatureServer/0` (13)	2026-04-08	Not stated	`parks`; access pillar
Regional trails (Allegheny County planning)	`COUNTY/Regional_Trails_(PUBLIC)/FeatureServer/0` (3,193 segments; 3,192 on the map)	2025-01-02	Not stated	`trails`
City playgrounds (City of Pittsburgh, via WPRDC)	data.wprdc.org (resource `47350364-44a8-4d15-b6e0-5f79ddff9367`, 125)	2022-12	CC BY	`places-playgrounds`
Citiparks pools and recreation centers (City of Pittsburgh)	`CITY/Citipark_Facilities/FeatureServer/1` (rec centers), `/2` (pools)	2022–2026	City open data	`places-citiparks`
Registered Community Organizations (City of Pittsburgh)	`CITY/PGHWebRCO/FeatureServer/0` (45; contact fields dropped)	edited 2026-08-05	City open data	`rcos`
Pittsburgh Zoning Code, Title Nine, incl. §911.02 use table (City of Pittsburgh, published by General Code eCode360)	ecode360.com (§911.02); print views `https://ecode360.com/print/PI6865?guid=` 45475424 (Art. IV), 45476514 (Art. V), 45479638 (Art. IX); TOC ecode360.com and ecode360.com ; other sections read directly (e.g. ecode360.com , ecode360.com , ecode360.com); SP-10 Appendix PDF (eCode360 attachment 336977)	legislation through 2026-09-16	Code of record	`typology-district-matrix.csv`, `special-district-residential-permissions.csv`, `use-definitions.csv`; `legal-pathway-plan`, `residential-zoning` layers; pillar legal multiplier
Zoning Code use classifications handout (City DCP)	pittsburghpa.gov	–	City document	Use-size thresholds for senior and care housing
Zoning overlays and historic districts (City of Pittsburgh)	`CITY/InclusionaryHousingOverlayDistrict/FeatureServer/0`, `CITY/PGHWebParkingReductionOverlay/FeatureServer/0`, `CITY/PGHWebMajorTransitBuffer/FeatureServer/0`, `CITY/Multi_Unit_Zoning_Districts/FeatureServer/0`, `CITY/PGHWebCHDHistoricDistricts/FeatureServer/0`, `CITY/PGHWebZoningOverlays/FeatureServer/0`	2026	City open data	`city-zoning-overlays` (608 features)
Suburban municipal zoning (each municipality's GIS; several listed in PRT's registry)	Penn Hills services5.arcgis.com · Bethel Park services7.arcgis.com · Monroeville services9.arcgis.com · Mt. Lebanon services8.arcgis.com · Whitehall services8.arcgis.com · Upper St. Clair services6.arcgis.com · Dormont services6.arcgis.com · Moon services8.arcgis.com · McCandless services1.arcgis.com	2026	Not stated (municipal)	`suburban-zoning` (5,872 polygons)
Zoning Board of Adjustment decisions (City of Pittsburgh DCP)	Decision PDFs under pittsburghpa.gov (1,362 PDF links), found via the Wayback Machine CDX API web.archive.org and the archived ZBA page (253 links); agenda page pittsburghpa.gov ; board page pittsburghpa.gov	2023–Aug 2026 (476 decisions)	Public records	`zba-decisions` layer and outcome tables
City Council land-use matters (Pittsburgh City Council, Legistar)	Legistar Web API webapi.legistar.com (284 matters, 2000–2026-09-26); pittsburgh.legistar.com ; attachments on pittsburgh.legistar1.com	2000–2026	Public records	`council-land-use-actions` (183 geocoded)
Planning Commission minutes 2020–2025 (City DCP)	pittsburghpa.gov (one PDF per year)	2020–2025	Public records	`planning-commission-actions-parsed.csv`
OneStopPGH permit records `OSPI_H` (City of Pittsburgh)	pghbridgis.pittsburghpa.gov	pulled 2026-09-26	City data	`permits-by-type` (1,088 new-housing permits, days to issue)
Development & Construction Projects v2 (City of Pittsburgh; legacy, stale)	`CITY/Development_Construction_Projects_v2/FeatureServer/0`	2020–2023 (edited 2023-09-11)	City open data	`permits-by-type` (144 legacy records)
Allegheny County parcels MapServer (Allegheny County GIS)	gisdata.alleghenycounty.us	current	County data	Parcel matching for Council actions
PA DHS Human Services Provider Directory; Licensed Personal Care Homes (PA DHS)	humanservices.dhs.pa.gov (Office of Long-Term Living, Allegheny); data.pa.gov	2026-09-22	Public	`senior-supportive-housing`; `care-facility-spacing` (800 ft rule)
511PA traffic cameras (PennDOT)	511pa.com (still images); camera list from the site's data endpoint 511pa.com (DataTables query parameter, 100 rows per page), which yields 210 PennDOT stills (image `/map/Cctv/<id>`, ~60 s) plus 8 PA Turnpike HLS feeds; video URLs via 511pa.com (PennDOT streams on pa-se1.arcadis-ivds.com returned 401, so stills are used)	—	511PA terms	`cameras`
PA Turnpike traffic cameras (Pennsylvania Turnpike Commission)	paturnpike.com ; camera API `https://www.paturnpike.com/traveling/traffic-cameras/getbyroadway?roadwayId=` (needs header `X-Requested-With: XMLHttpRequest`; in-county roadways I-76 `ef2829cb-fea1-4083-af4c-9bfbd85b8568` and PA-576 `2e8b2a93-1f81-4fed-8919-58b33b638eb8`; CloudFront streams are Referer-gated) (HLS streams on CloudFront, played through the app's same-origin proxy)	—	PA Turnpike terms	`cameras`
USGS HIVIS river cameras (USGS)	camera registry api.waterdata.usgs.gov (needs an `api-key` header; the public key embedded in the HIVIS web app is used and isn't reproduced here); 2 cameras in Allegheny; apps.usgs.gov ; apps.usgs.gov (images `https://usgs-nims-images.s3.amazonaws.com/720/<camId>/<camId>_newest.jpg`)	—	Public domain	`cameras`
Breathe Cam industrial smoke cameras (CMU CREATE Lab)	breathecam.org ; list breathecam.org ; embedded as `https://breathecam.org/#s=<id>` (braddock2, westmifflin2, accan2, cementcam)	—	CREATE Lab terms	`cameras`
WeatherSTEM Allegheny sky cameras (WeatherSTEM)	allegheny.weatherstem.com ; allegheny.weatherstem.com (station list allegheny.weatherstem.com ; snapshots `https://images.weatherstem.com/skycamera/allegheny/<handle>/cumulus/snapshot.jpg`; the `pitt` camera has been frozen since 2024 and is excluded)	—	WeatherSTEM terms	`cameras`
EarthCam: Pittsburgh (Troy Hill) and The Andy Warhol Museum FigmentCam / ChurchCam	earthcam.com ; warhol.org ; earthcam.com	—	EarthCam terms; re-streaming is forbidden, so the pages are embedded as sandboxed iframes	`cameras`
PixCams bald eagle and red-tailed hawk nest cams (U. S. Steel, Duquesne Light; YouTube embeds)	pixcams.com and related pages; 7 YouTube live embeds: U. S. Steel eagles `2CP8QA_xKx4`, `9njkml1oQRQ`, `uxzRXhKshow`, `a83Y5shllLg`, `dDIfhPXn_-c`; hawk `jUS-CaVPNYU`; Hays eagle `yPq9aNukckQ`	—	Owner terms	`cameras`
Pittsburgh Zoo & Aquarium penguin and cheetah cams (Ozolio)	pittsburghzoo.org ; `.../cheetah-webcam/` (Ozolio embeds `EMB_QVKU00000674` penguin, `EMB_ODOJ00000679` cheetah)	—	Owner terms	`cameras`
Discover the Burgh North Side skyline cam (Nest)	discovertheburgh.com	—	Owner terms	`cameras`
PPG Place plaza cam	ppgplace.com ; StarDot image 96.69.79.178 (HTTP, served through the app proxy)	—	Owner terms	`cameras`
W3SLL personal weather station cam (Squirrel Hill)	wx.w3sll.net ; image wx.w3sll.net (HTTP, served through the app proxy)	—	Owner terms	`cameras`
Camp Guyasuta cam (WeatherBug, via the Webcam Galore cache)	image images.webcamgalore.com (~30 min refresh); its former page webcamgalore.com returned 404 on 2026-09-27	—	Owner terms	`cameras`
Automatic license plate reader locations (OpenStreetMap contributors, incl. DeFlock mapping)	Overpass extract saved as `inputs/cameras/alpr.json` (956 OSM features; ~225 Flock Safety). Query: `nwr["man_made"="surveillance"]` and `nwr["surveillance:type"]` in bbox 40.19,-80.37,40.68,-79.68, `out center tags`, clipped to the TIGERweb county polygon (GEOID 42003). Tags carry manufacturer, operator and direction; deflock.me	—	ODbL	`alpr-cameras` (locations only, no feeds)

§10 · licenses

Licenses, attribution and runtime services

Attribution obligations

Source	License / terms	What we must do
OpenStreetMap (basemap tiles, 3D buildings in CARTO tiles, Overpass extracts, Nominatim search, ALPR locations, post offices, laundromats, dentists, community centers, police/EMS additions)	ODbL 1.0	Show "© OpenStreetMap contributors" with a link to openstreetmap.org. Share-alike applies to derived databases we redistribute. Respect the tile and Nominatim usage policies (low volume, identifying User-Agent).
CARTO basemaps (Dark Matter, Positron)	CARTO basemap terms	Show "© CARTO" next to the OSM credit.
Mapping Inequality (HOLC 1937 redlining)	CC BY-SA 4.0	Credit "Mapping Inequality, Nelson et al., University of Richmond Digital Scholarship Lab". Derived redistribution must stay CC BY-SA.
UMN Accessibility Observatory, Access Across America: Transit 2024	CC BY-NC 4.0	Credit UMN; non-commercial use only. Fine for the hackathon; a commercial pilot would need permission.
DOE LEAD Tool 2022	CC BY 4.0	Credit "U.S. DOE Low-Income Energy Affordability Data (LEAD) Tool, 2022 update".
WPRDC datasets	CC0 or CC BY (varies by dataset; listed per row below)	For CC BY rows, credit the publishing agency and WPRDC.
Zillow Research (ZORI)	Zillow Research data terms	Credit "Zillow Research". Aggregates only.
Child Opportunity Index 3.0	diversitydatakids.org terms (free with attribution)	Credit diversitydatakids.org. Note that we read a third-party ArcGIS mirror.
Princeton Eviction Lab (Eviction Tracking System)	Eviction Lab data terms	Credit "The Eviction Lab at Princeton University".
Live camera feeds	Each owner's terms; feeds are embedded or proxied, never re-hosted	Credit the operator in the viewer (the app does this per camera).
Everything else in this document	U.S. federal public domain, or state/county/city open data with no stated license	Credit the publisher. Where "not stated" appears, no license string was published.

AI models and runtime services

Service	How it is used	Endpoint
Jev ("TypeSafe") decision model, served through OpenRouter	Site-fit scores per housing type (with confidence) shown in the typology panel and cited by the chat companion; outputs are labeled "assumption", not evidence	openrouter.ai (model set by the `SYSTEM_ONE_MODEL` environment variable). About Jev: typesafe.ai ; dev.to ; press: marktechpost.com , tomshardware.com
Google Gemini API	Chat companion (tool calling over the scoring model) and read-aloud voice	generativelanguage.googleapis.com
Neon Postgres + PostGIS	Hosts the ~585k county parcels for viewport queries	Neon (serverless Postgres)
Microsoft Planetary Computer	STAC search and pixel extraction for Landsat surface temperature	planetarycomputer.microsoft.com ; `/api/data/v1`
Census Reporter API	ACS tables and census geometries without an API key	api.censusreporter.org
Overpass API	OpenStreetMap extracts	overpass-api.de ; overpass.kumi.systems
Socrata catalog API	Discovering data.pa.gov and data.cdc.gov datasets	api.us.socrata.com
ArcGIS Online search	Discovering public feature services	arcgis.com

Legal-feasibility datasets built by the team

City of Pittsburgh · raw coded facts, no scores

File	Primary sources
`care-facility-spacing-800ft.geojson`	PA DHS / DOH facility records; Zoning Code §911.04 spacing rule (800 ft)
`council-land-use-actions.csv`	Legistar Web API (webapi.legistar.com/v1/pittsburgh/matters), Allegheny County parcels MapServer, Census geocoder
`council-land-use-actions.geojson`	Legistar Web API (webapi.legistar.com/v1/pittsburgh/matters), Allegheny County parcels MapServer, Census geocoder
`pathways.csv`	Zoning Code Ch. 922 procedures (eCode360)
`permits-by-typology-district.csv`	Derived from permits-new-residential
`permits-new-residential.geojson`	OneStopPGH OSPI_H, WPRDC PLI permits, City Development_Construction_Projects_v2, PGHWebZoning, PGHWebNeighborhoods
`senior-and-group-housing.geojson`	PA DHS provider directory, data.pa.gov personal care homes and nursing homes, CMS nursing home provider info, HUD MF-Assisted / 202 / 811 / LIHTC / Public Housing, HACP communities, City boundary, PGHWebZoning, Census geocoder
`typology-district-matrix-wide.csv`	Zoning Code §911.02 (eCode360), live distinct `zon_new` query on PGHWebZoning
`typology-district-matrix.csv`	Zoning Code §911.02 (eCode360), live distinct `zon_new` query on PGHWebZoning
`zba-decisions.csv`	ZBA decision PDFs on pittsburghpa.gov, located via the Wayback Machine CDX API and the archived ZBA page; geocoded
`zba-decisions.geojson`	ZBA decision PDFs on pittsburghpa.gov, located via the Wayback Machine CDX API and the archived ZBA page; geocoded
`zba-outcomes-by-district-typology.csv`	Derived from zba-decisions

§11 · assumptions

Assumptions register

Everything we chose rather than measured, in one place. If you disagree with one, the weights popover or the config file is where to change it.

Method assumptions

18 items

Area	Assumption	Where
Scoring	Missing data is excluded and the remaining weights renormalized; it is never filled with a guess at indicator level.	score.ts · E3
Scoring	A pillar with too little data counts as that pillar's City 25th-percentile score, so missing data is never rewarded.	pillars.config.json · impute
Scoring	Pillar scores are floored at 1 inside the geometric mean so one zero doesn't zero the overall.	overall.floor
Scoring	All pillar weights default to 1. Equal weights are a starting point, not a finding; presets show other value sets.	pillars[].weight
Zoning	The overall zoning factor uses the easiest pathway among the five mainstream housing types unless the user picks one type in the panel; senior and group housing are scored separately in the typology panel.	legal.typologies · legalLevel
Scoring	Deal-killer hazards (floodway, sliver lot, mapped mines) multiply the overall score; the lowest factor applies. Mines get a mild ×0.9 because about 30% of City parcels are over mapped mines and the layer has no depth of cover.	overall.hazard_multiplier
Verdict	The worst reason sets the verdict color; unknown outranks green so missing data never reads as buildable. Mines are yellow (investigation needed), not red.	verdict.ts
Verdict	A lot is flagged for a likely setback variance when its width minus the §903.03 interior side setbacks leaves under 14 ft. Width is the short side of the oriented bounding rectangle; corner-lot exterior setbacks are not modeled.	verdict.side_setbacks
Pencil	Houses are valued at the tract's median resale price (close to an appraiser's comps); 2+ units at capitalized block-group rent. Both describe existing stock and understate new-build value, so the check leans pessimistic.	pencil.value_bias
Pencil	Soft costs 15%, margin 10%, rent-to-value ×103 (40% operating costs, 7% cap rate) and the 35% subsidy floor are our assumptions; construction $/sf and site cost come from hackathon experts and are editable.	pencil.*
Zoning	A parcel within 30 m of a district that allows attached or multi-unit homes is treated as a border case (×0.35 instead of ×0.2).	legal.border_m
Zoning	Where a district has no local Zoning Board cases, the approval likelihood is a neutral 0.7.	typology-panel.tsx · E9
Zoning	Rezoning closeness uses a simple density ladder R1D → R1A → R2 → R3 → RM; non-residential districts count as distant (0.2).	typology-panel.tsx · E9
Zoning	Rules are a simplified transcription of §911.02 / §903.03 (post-May-2025 minimum lot sizes). Anything not encoded shows as unknown, never guessed.	site-fit.ts
Site fit	Lot width and depth come from the parcel's minimum oriented bounding rectangle, which overstates buildable area on irregular lots.	parcels.ts · typologyFit
Site fit	Hazard shares under 1% of the lot are not mentioned to the model.	site-fit.ts · buildSiteState
Site fit	Ratings with confidence below 0.3 (House: 0.2) are shown but flagged for review; the thresholds were tuned on observed answers (~0.3–0.55).	site-fit.ts · REVIEW_CONFIDENCE
Data	Area indicators (tract, block group, ZIP) are assigned to every parcel inside that area. They describe the neighborhood, not the lot.	every area indicator

Indicators labeled assumption or value judgment

3 indicators

Indicator	Evidence	Weight	Why
New homes permitted within 800 m, 2019–26 (log units) demand_new_construction	assumption	1.5	Revealed demand: builders are building nearby. Counts homes, not permits, on a log scale so one 224-unit building counts more than one house but doesn't swamp everything, and includes apartment buildings, which the City files as 'Commercial'. Unit counts are parsed from permit text (assumption where missing: 1). Several permits on one parcel (phases, trades) count once, at the largest unit count.
Market Value Analysis type (A strongest … J most distressed) demand_market_strength	value	0	Default weight 0: rewarding strong markets steers housing away from distressed neighborhoods. Available for users who choose to weigh it.
Median gross rent vs. rent affordable at 50% AMI (3-person household) afford_rent_vs_ami	assumption	0	Context only (weight 0): rent relative to AMI measures how expensive the market is, not how many households are struggling (round-1 review). 1242.5 = 30% × $49,700 (HUD FY2026 Pittsburgh 50% AMI, 3 persons) ÷ 12.

Shown for context, weight 0 (not scored)

14 indicators

Rent growth, last 12 months (Zillow ZORI)Market Value Analysis type (A strongest … J most distressed)Share of lot in the FEMA regulatory floodwayKnown lead or galvanized service line at this addressFEMA NRI inland flooding scoreMedian home sale price, 2024–25 (census tract)Easiest legal pathway for mainstream housing types (current code)Site availability (what is on the parcel now)Lot area (for the sliver-lot check)Median gross rent vs. rent affordable at 50% AMI (3-person household)Housing + transportation cost share, median-income family (HUD LAI v3)Eviction filings per 100 renter households, 2025Residents enrolled in college (ACS 2020–24)FEMA NRI heat wave score

Data traps we found and handled

verified during the build

ACHD's food-facility "open" list still includes closed businesses (Blockbuster, Ames, Eckerd). A facility is kept only if it was inspected in the last ~2 years; test accounts are filtered out.
Rite Aid closed every store by Sept 2025 but still appears in NPPES under Rite Aid, Thrift Drug and Eckerd; these are removed.
USDA SNAP retailer records repeat some stores; they are deduplicated.
ACHD blood-lead 2021–24 columns have no published definition and run 5–10× the county rate; only pooled 2015–20 is used.
County sales for 2018–19 have far fewer "valid sale" codes than later years; 2020 is the baseline.
City police incident-level data ends in Nov 2023; crime is not used in any score.
Census tracts starting 98xxxx are special-use (parks, airport); percent changes there are blanked below ~100 households.
ZIP-based filters leak across the county line, so every layer is clipped to the county polygon.
DOE LEAD tenure codes are REN/OWN, not RENT.
Pittsburgh has been a mandatory HUD Small Area FMR area since 2018-04-01; SAFMR, not metro FMR, is the rent benchmark.

§12 · limitations

What this tool can't tell you

Not legal advice

Zoning is a simplified interpretation of the City code. Verify with the Zoning Administrator before acting.

City of Pittsburgh zoning only

Yinzone covers City of Pittsburgh parcels only. Other municipalities, including the Mount Oliver enclave, have their own zoning codes, which are not encoded: parcels there show no scores, verdicts or zoning, and the map greys them out. Some context layers still extend across the county.

Only a rough cost check

The pencil check compares practitioner cost ranges with nearby sale prices and rents. Land, financing, subsidy terms and new-build premiums are not modeled; it is a first screen, not a pro forma.

No project-level approval odds

Zoning Board rates describe districts, not your application.

Excluded on purpose

Crime and race never enter any score. Income and rent are context for need, not a filter.

Not covered

Water and sewer capacity, school quality, and tornado risk are not scored.

Proxies

Air quality is an emissions and modeled-exposure proxy (EJScreen), not monitored air at the lot.

Area averages

Tract, block-group and ZIP indicators describe the neighborhood, not the parcel.

Vintages differ

Sources range from 2015–20 pooled health data to live feeds; each layer shows its own as-of date.

AI is advisory

Jev site fit is a model judgment with shown confidence. The chat only explains cited facts and can't see other map layers.

§13 · references

Methodology references and sources not used

Composite-indicator standards and index designs we followed

61 references

[01]oecd.org 9789264043466-en.pdf
[02]knowledge4policy.ec.europa.eu 10-step-pocket-guide-to-composite-indicators-and-scoreboards
[03]wipo.int wipo-pub-2000-2023-appendix3-en-â-appendix-ii-full-version-â
[04]hdr.undp.org hdr2023-24_technical_notes.pdf
[05]hdr.undp.org hdrp201101.pdf
[06]oecdbetterlifeindex.org better-life-initiative
[07]web.archive.org better-life-initiative
[08]oecd.org oecd-better-life-index.html
[09]livabilityindex.aarp.org faqs
[10]livabilityindex.aarp.org methods-sources
[11]treasurer.ca.gov CTCAC-HCD-2025.pdf
[12]eclass.hua.gr 12.GIS based multicriteria decision analysis a survey of the
[13]vpp.sbuf.se How to make decision AHP.pdf
[14]en.wikipedia.org TOPSIS
[15]en.wikipedia.org PROMETHEE
[16]asprs.org 1995_may_539-547.pdf
[17]ideas.repec.org v41y2009i12p1513-1523.html
[18]climateprogramportal.org cejst-technical-support-document.pdf
[19]oehha.ca.gov calenviroscreen40reportf2021.pdf
[20]nationalacademies.org 8
[21]ideas.repec.org v168y2005i2p307-323.html
[22]arxiv.org 1104.3009
[23]smaa.fi tervonen-phd.pdf
[24]pmc.ncbi.nlm.nih.gov PMC10986280
[25]huduser.gov PittsburghPA-CHMA-24.pdf
[26]alleghenycounty.us 2021-mva-executive-summary.pdf
[27]reinvestment.com market-value-analysis
[28]data.wprdc.org market-value-analysis-2021
[29]zillow.com market-heat-index-methodology-34057
[30]yorku.ca malczewski00.pdf
[31]oregon.public.law oar_660-008-0005
[32]california.public.law government_code_section_65583.2
[33]leginfo.legislature.ca.gov codes_displaySection.xhtml?lawCode=GOV&sectionNum=65583.2
[34]law.cornell.edu 60.3
[35]huduser.gov bg_chas.html
[36]htaindex.cnt.org method-2022.pdf
[37]portland.gov download
[38]portland.gov gentrification_displacement_typology_analysis_2018_10222018.
[39]github.com udp_replication_project_methodology_10.16.2020-converted.pdf
[40]github.com displacement-typologies
[41]reinvestment.com measuring-displacement-risk-in-gentrifying-neighborhoods
[42]treasurer.ca.gov opportunity-map-methodology.pdf
[43]treasurer.ca.gov CTCAC-HCD-2026-Opportunity-Map-Memo-10-20-25.pdf
[44]docs.huduser.gov AFFH-T-Data-Documentation-AFFHT0007-August-2024.pdf
[45]docs.huduser.gov AFFH-T-Data-Documentation-AFFHT0004-November-2017.pdf
[46]ris.utwente.nl Geurs-accessibility.pdf
[47]pmc.ncbi.nlm.nih.gov PMC3547595
[48]walkscore.com methodology.shtml
[49]walkscore.com transit-score-methodology.shtml
[50]epa.gov national_walkability_index_methodology_and_user_guide_june20
[51]diversitydatakids.org COI 3.0 Technical Documentation 20260803_0.pdf
[52]atsdr.cdc.gov EJI_2024_Technical_Documentation.pdf
[53]fema.gov fema_national-risk-index_methodology-hazards-overview.pdf
[54]missionknox.org Appendix_4.3-FEMA_Scoring_Matrix-1.pdf
[55]fema.gov fema_national-risk-index_technical-documentation.pdf
[56]law.cornell.edu appendix-G_to_part_58
[57]airnow.gov technical-assistance-document-for-reporting-the-daily-aqi
[58]tandfonline.com 01944361003766766
[59]lincolninst.edu 2038_1360_LP2010-ch07-Land-Use-and-Vehicle-Miles-of-Travel-i
[60]pmc.ncbi.nlm.nih.gov PMC2801183
[61]carbonleadershipforum.org the-embodied-carbon-benchmark-report

Researched but not used, and why

32 sources

Source	Why it isn't used
PWSA sewersheds with combined-sewer overflow volume (`PWSA/Sewersheds/FeatureServer/0`, 485 polygons)	Researched as a sewer-stress proxy; the map shows sewer lines instead
ALCOSAN interceptors and outfalls (`https://services8.arcgis.com/vXrmg6iOl1pzNz4z/arcgis/rest/services/ALCOSAN_Collection_Network_Assets_LOCKED_Public/FeatureServer/4`, `ALCOSAN_Outfalls_LOCKED/FeatureServer/2`)	Regional context only
County sewer and water authority boundaries (`COUNTY/Allegheny_Sewer`, `COUNTY/Allegheny_Water`)	Context only
County "2100 projections for SFHA" (`COUNTY/2100_Projections_for_SFHA_NEW/FeatureServer/0`)	Offered as an optional future-flood layer; not integrated
City residential lot dimensions (`CITY/Parcels_Exp_02052025_Residential_Lot_Dimensions/FeatureServer/0`) and ETHOS lot suitability (`CITY/ETHOS_Lot_Suitability/FeatureServer/0`)	Provided to the pillar work; ETHOS scores are City composites, used only as a cross-check
County business district streets (`CITY/Business_District_Streets/FeatureServer/0`)	Optional; not integrated
MSI "possible mining" zones (PA DEP `MSI_SubsidenceRiskMiningPossible`)	Only the "confirmed" zones are mapped
CDC PLACES social-needs measures (housing insecurity, utility shutoff, transportation)	Empty for Pennsylvania in the 2024 and 2025 releases (stated as a data gap)
ACHD 2021–2024 blood-lead columns	No published definition; inconsistent with ACHD's own county rate (~3.1% in 2023)
U.S. DOT BTS National Transportation Noise Map (`maps.bts.dot.gov`)	Service unreachable on 2026-09-26
County WIC offices (`COUNTY/WICOffices`)	No usable names
Postal_Service_Vacant_Parcels (City; a Regrid export)	Contains owner names and mailing addresses; excluded for privacy and licensing
Homeless and domestic-violence shelter locations	Deliberately excluded (DV shelter locations are confidential)
Allegheny County DHS landlord/tenant dashboard (tableau.alleghenycounty.us)	Tableau view only; no downloadable tract data
FBI Crime Data Explorer	Needs an API key and agency-to-municipality matching
WPRDC air quality: monitor readings, sensor locations, forecasts, inversions, 2011 PM2.5 (data.wprdc.org , resource `b646336a-deb4-4075-aee4-c5d28d88c426`; data.wprdc.org ; data.wprdc.org ; data.wprdc.org)	Monitors too sparse to vary by parcel; forecasts aren't site characteristics; 2011 PM2.5 superseded by EJScreen
EPA facility points (EMEF, geopub.epa.gov) and PA DEP eMapPA extraction (gis.dep.pa.gov)	Probed during integration; the ACHD inventory was used instead
WPRDC parcel-level lead hazard file (resource `2ddfd798-b71a-4f78-bc17-8c54c6a30511`)	The map uses PWSA's newer address points instead
City individual historic properties (`CITY/PGHWEBCHDIndividialProperties/FeatureServer/0`)	Researched for the legal work; historic districts are mapped instead
City parks `PGHWebParks`, TPL ParkServe priority areas (`CITY/TPL_ParkServe_ParkPriorityAreas/FeatureServer/0`), county park hiking trails (`COUNTY/Allegheny_County_Trails/FeatureServer/0`)	County-wide park layers were used instead
County municipal boundaries (`COUNTY/Allegheny_County_Municipal_Boundaries/FeatureServer/0`), ConnectGovs fire districts (`CG/Allegheny_County_Fire_Districts/FeatureServer/0`)	Context only
City `Police_Incidents_Blotter` layer	Stale (2023); the WPRDC archive was used
HUD CHAS on ArcGIS (`HUD/ACS_5YR_ESTIMATES_CHAS_TRACT/FeatureServer/1`) and the 2017–21 dictionary	2013–2017 vintage; the 2018–2022 files were used
Esri Landsat Level-2 ImageServer (landsat.imagery1.arcgis.com, utility.arcgis.com proxy)	Requires a token; Microsoft Planetary Computer was used instead
OpenFreeMap tiles (tiles.openfreemap.org)	Recommended in early research; the app uses CARTO Positron / Dark Matter and OSM raster tiles instead
Census Bureau API (api.census.gov)	Keyless requests are redirected to its missing-key page (checked 2026-09-27); Census Reporter was used instead
Municode (library.municode.com ; api.municode.com)	The City's zoning GIS links each district to its eCode360 section, so the legal work used eCode360; the Municode library is a JavaScript app that wasn't scraped
Allegheny County Housing Needs Assessment (alleghenycounty.us) and the City of Pittsburgh Housing Needs Assessment (January 2022; pittsburghpa.gov ; implementation engage.pittsburghpa.gov)	The County assessment is listed in the organizer data catalog; both are too aggregated for parcel scoring and are used as policy context
CDC PLACES 2025 release (data.cdc.gov `yjkw-uj5s` GIS format, `cwsq-ngmh` long format)	For Pennsylvania tracts it carries only 5 measures (dental visits, teeth lost, short sleep, colon and mammography screening; checked 2026-09-27), none of the health outcomes the layer shows, so the 2024 release is used
Water-lead context pages cited in answers: pgh2o.com , gettheleadoutpgh.org , amwater.com , alcosan.org	Background for the lead and sewer layers
Allegheny County eviction analyses: analytics.alleghenycounty.us ; analytics.alleghenycounty.us	Context; no tract-level download
Legal references consulted by the legal-feasibility work: PA Municipalities Planning Code (dep.state.pa.us); Babst Calland, "The devil is in the details and the deemed approval deadlines" (babstcalland.com)	Background for deemed-approval and hearing timelines

Source manifest: DATA_SOURCES.md (low-level data session, compiled 2026-09-27).

Base district	Cases	Approved	Denied	Approval rate
R1D	101	87	11	86%
R1A	83	74	7	89%
R2	62	46	13	74%
LNC	49	45	3	92%
RM	31	30	1	97%
UI	23	20	2	87%
UNC	17	12	3	71%
EMI	13	11	1	85%
H	12	10	1	83%

Base district	Cases	Approved	Denied	Approval rate
R1D	101	87	11	86%
R1A	83	74	7	89%
R2	62	46	13	74%
LNC	49	45	3	92%
RM	31	30	1	97%
UI	23	20	2	87%
UNC	17	12	3	71%
EMI	13	11	1	85%
H	12	10	1	83%

Base district	Cases	Approved	Denied	Approval rate
R1D	101	87	11	86%
R1A	83	74	7	89%
R2	62	46	13	74%
LNC	49	45	3	92%
RM	31	30	1	97%
UI	23	20	2	87%
UNC	17	12	3	71%
EMI	13	11	1	85%
H	12	10	1	83%