Methodology

Every number on the board traces to a public government dataset. This page documents what each category measures, where it comes from, how it was computed and what to watch out for.

Geography

The units are the Department of City Planning's 2020 neighborhood tabulation areas (NTAs) — 262 areas built from census tracts. The board ranks the 197 residential NTAs (City Planning type 0) and excludes parks, cemeteries, airports and Rikers Island, which have few or no residents. Boundaries come from the 2020 NTA file on NYC Open Data, which is clipped to the shoreline; land areas are computed from those boundaries (square feet ÷ 27,878,400).

Census categories

All demographic, economic, housing and commuting categories come from the American Community Survey (ACS) 2020-2024 five-year estimates, as tabulated at the NTA level by the Department of City Planning and published as four Excel workbooks (demographic, social, economic, housing) on City Planning's ACS page. Values were read directly from those files with no adjustment. Category definitions (universes such as "population 25 and older") are noted under the board title.

Margins of error. ACS figures are survey estimates. The board shows the Census Bureau's published 90% margin of error (±) next to each estimate where one exists. Neighborhoods whose values sit within each other's margins are statistically indistinguishable; treat near ties as ties.
Capped medians. City Planning computes neighborhood medians from bracketed distributions whose top brackets are open-ended ($200,000-or-more income, $3,500-or-more rent, $2,000,000-or-more home value), so a median that falls in the top bracket is reported at the cap. Values at the cap display with a plus sign and tie for first.
Suppression. A handful of values are unpublished for small neighborhoods; those rows show "no data" and are not ranked in that category.
Derived categories. "Share never married" = never-married men plus never-married women, divided by population 15 and older. "Single-family homes" = 1-unit detached share plus 1-unit attached share. Population density = ACS population ÷ land area.

Administrative counts and rates

Event data was downloaded from NYC Open Data and the state open data portal, geocoded to NTAs by point-in-polygon assignment against the 2020 boundaries (99%+ of records matched in every dataset), then converted to rates using ACS denominators.

Category	Dataset	Window	Notes
Felony complaints; violent felonies	NYPD complaint data (historic)	Calendar 2025, by report date (RPT_DT)	Violent = murder and non-negligent manslaughter, rape, robbery, felony assault. Complaints are counted where the crime occurred; job centers like Midtown rank high relative to their resident population.
311 requests; noise; rodent; illegal parking	311 service requests	Aug. 1, 2025-July 31, 2026	Noise = every complaint type beginning with "Noise." 311 measures reporting, not conditions; a few prolific callers can move a small neighborhood's rate.
Traffic crashes; crash injuries	Motor vehicle collisions	Calendar 2025	Police-reported crashes only. Injuries include deaths. About 3% of crashes lack coordinates and are excluded.
Evictions	Evictions (Department of Investigation)	Executed in calendar 2025	Residential evictions executed by city marshals, per 1,000 renter-occupied units (ACS). Undercounts total displacement, which mostly happens before a marshal arrives.
Restaurants	Restaurant inspection results	Current active file	Unique establishments (by CAMIS id) in the Health Department's inspection file.
Parks-managed trees	Parks Department forestry tree points	Current	Standing live trees only: points with structure "Full" (which excludes the roughly 174,000 "Retired" records for removed trees, and stumps) and condition other than "Dead." Covers street trees plus park trees the agency actively manages; natural forest isn't inventoried tree by tree. 888,653 trees citywide.
Subway stations	MTA subway stations	Current	Stations (not complexes) whose coordinates fall inside the boundary — Times Square-42nd Street counts as five stations — including the 21 Staten Island Railway stations. 496 systemwide. Stations just outside a neighborhood's edge don't count toward it.

Make-your-own combinations

The "Mix your own" tab builds two kinds of user-defined rankings, computed in the browser from the same category values shown on the board:

Ratio. One category's value divided by another's, for each neighborhood. A neighborhood shows "no data" if either component is unpublished or the denominator is zero. Ratios of top-coded medians (income, rent, home value) use the capped value, so extreme neighborhoods are conservative. Rank 1 is the highest ratio.
Blend. The average of a neighborhood's ranks across 2-6 chosen categories, weighted equally, where rank 1 in a category means the highest value. A neighborhood must have data in every chosen category to be ranked. Rank 1 is the lowest average.

Combinations are arithmetic on published figures, chosen by the reader; they are not editorial indexes, and margins of error are not propagated through them. Shareable links (the page address updates as you build) reproduce the same combination.

Ranking rules

Ranks run 1-197 among neighborhoods with published data; ties share a rank.
Bars are proportional to the value and always start at zero.
Borough filters and search change what is displayed but never change a neighborhood's rank, which is always computed against all 197.
No category is labeled "best" or "worst." High felony rates and high restaurant counts are both just facts about a place; the reader decides what they mean.

Honest limitations

Per-resident rates use nighttime (residential) population. Neighborhoods with huge daytime populations — Midtown, the Financial District — will look extreme on anything divided by residents.
ACS five-year estimates average 2020-2024; fast-changing neighborhoods lag reality. The 2020-2024 release (published December 2025) is the newest neighborhood-level data the Census Bureau produces. Fresher population figures exist only citywide: the bureau's Vintage 2025 estimates put New York City at about 8.58 million as of July 1, 2025 — roughly 1.2% above the five-year ACS total of 8.48 million, which centers on 2022, near the pandemic population trough. The 197 neighborhoods here sum to 8,475,754; the small gap to the citywide ACS total is people living in excluded park, airport and institutional areas.
311 and crime complaints measure what gets reported, which varies with reporting habits, not only conditions.
The eviction, restaurant and tree files are administrative and inherit their agencies' lags and gaps.

Reproducibility

The full pipeline — fetch scripts, point-in-polygon aggregation and the build step that produces rankings.json — is in the GitHub repository. Data was pulled Aug. 2, 2026. This project was built with the help of artificial intelligence (Claude); if a number looks wrong, check it against the source datasets linked above.

← Back to the board