NYC Open Data field reference

Dataset ids, query conventions and field names for the eleven NYC Open Data (Socrata) datasets Complied ingests, with how each field lands in public_events or buildings. For engineers writing or debugging a feed; architecture and scheduling are in ingestion. Column lists were checked against each dataset's live metadata.

Query conventions

Every dataset is a Socrata endpoint:

https://data.cityofnewyork.us/resource/<DATASET_ID>.json?<SoQL>

• Page with $limit (50,000 in the pipeline), $offset and a stable $order (:id, or a unique column). Loop until a page is shorter than the limit.
• $where, $select and $order are the filter, projection and sort. count(*) ($select=count(*) as n) sizes a backfill window.
• Send the app token as X-App-Token or $$app_token (NYC_OPEN_DATA_TOKEN) for higher rate limits.
• The hidden :updated_at system field supports incremental pulls where the publisher updates a few rows per night: $where=:updated_at > '<floating timestamp>'. It is only usable when each nightly stamp touches only the rows that changed; check with $select=:updated_at, count(*) … $group=:updated_at. DOB complaints is replaced wholesale each night (every row shares one stamp), so it uses a status-and-dates filter instead. Only small datasets are pulled in full.
• Date shapes differ. timestamp columns are floating timestamps (2020-01-01T00:00:00). DOB violations and ECB use compact YYYYMMDD text, and the literal string "0" (or an impossible date such as month 20) means "no date". DOB complaints uses MM/DD/YYYY text, which does not sort, so no >= range works on it. sanitizeCompactDate, sanitizeMdyDate, toDayString and parseDay in feeds/dates.ts parse them, and mdyRangeClause in feeds/socrata.ts writes the MM/DD/YYYY backfill windows.
• Every normalized event carries agency, event_type, a source_id prefixed by the feed, status_norm (OPEN, CLOSED, DISMISSED), and an optional status_detail in the agency's own vocabulary. The whole source row is kept only for quarantined rows (unlinked_events.raw_data); public_events keeps no copy of it: the Source record panel fetches the row live from NYC Open Data. HPD violations, ECB and OATH also fill promoted columns (class, certify-by, penalty, balance, hearing result and so on); which source field feeds each is in ingestion: promoted fields.

Feed	Dataset	Id	agency	source_id	Update mode
hpd-violations	Housing Maintenance Code Violations	wvxf-dwi5	hpd	hpd-violation:<violationid>	Incremental
hpd-complaints	HMC Complaints and Problems	ygpa-z7cr	hpd	hpd-complaint:<complaint_id>	Incremental
hpd-litigation	Housing Litigations	59kj-x8nc	hpd	hpd-litigation:<litigationid>	Full pull
dob-violations	DOB Violations	3h2n-5cm9	dob	dob-violation:<number>	Incremental
dob-complaints	DOB Complaints Received	eabe-havv	dob	dob-complaint:<complaint_number>	Active + recent dates
ecb-violations	DOB ECB Violations	6bgk-3dad	ecb_oath	ecb-violation:<ecb_violation_number>	Incremental
oath-hearings	OATH Hearings Division Case Status	jz4z-kudi	fdny, dsny, dohmh	oath-hearing:<ticket_number>	Incremental
nyc-311	311 Service Requests from 2020 to Present	erm2-nwe9	dep, dsny, fdny, dot	311:<unique_key>	Incremental
dof-liens	Tax Lien Sale Lists	9rz4-mjek	dof	dof-lien:<bbl>	Weekly full pull
Layer 1	BUILDING (footprints)	5zhs-2jue			Manual
Layer 1	PLUTO	64uk-42ks			Manual

The enforcement event feeds and the record feeds (see ingestion: record feeds). The event feeds run in the nightly delta and in the weekly full sync; the record feeds only in the weekly full sync (ingest records). ingest sample loads the newest N rows of any of them. The building key is what resolution uses; most also carry an address.

Feed	Dataset	Id	Writes	source_id	Building key
hpd-vacate-orders	Order to Repair/Vacate Orders	tb8q-a3ar	event vacate_order	hpd-vacate:<vacate_order_number>	bin, bbl
hpd-aep	Buildings Selected for the AEP	hcir-3275	event aep_selection	hpd-aep:<building_id>:<aep_round>	bin, bbl
hpd-omo-charges	Open Market Order (OMO) Charges	mdbu-nrqn	event emergency_repair_omo	hpd-omo:<omoid>	bin, bbl
hpd-hwo-charges	Handyman Work Order (HWO) Charges	sbnd-xujn	event emergency_repair_hwo	hpd-hwo:<hwoid>	bin, bbl
dob-safety-violations	DOB Safety Violations	855j-jady	event dob_safety:<program>; rows carried over from BIS (number starts with a MMDDYY date) skipped, dob-violations has them	dob-safety-violation:<violation_number>	bin, bbl
fdny-vacates	NYC Fire Department Building Vacate List	n5xc-7jfa	event fdny_vacate	fdny-vacate:<bin>:<vac_date>	bin, bbl
dohmh-rodent-inspections	Rodent Inspection	p937-wjvj	event rodent_inspection, one per job	dohmh-rodent:<job_id> (a row with no job: <bbl>:<day>:<inspection_type>)	bin; bbl, else from boro_code/block/lot
dot-sidewalk-violations	DOT sidewalk violations	6kbp-uz6m	event sidewalk_violation	dot-sidewalk:<violationid>	BBL only: bblid looked up in DOT's lot table i642-2fxq
hpd-registrations	Multiple Dwelling Registrations	tesw-yqqr	filing hpd_registration	hpd-registration:<registrationid>:<buildingid>	bin; BBL from boroid/block/lot
hpd-registration-contacts	Registration Contacts	feu5-w2e2	party (staff only)	hpd-contact:<registrationcontactid>	none: registrationid
hpd-bedbug-reports	Bedbug Reporting	wz6d-d3jb	filing hpd_bedbug_report	hpd-bedbug:<building_id> (latest period)	bin, bbl
hpd-conh	CONH Pilot Building List	bzxi-2tsw	designation conh_pilot	hpd-conh:<building_id>	bin, bbl
dob-elevators	DOB NOW Elevator Safety Compliance	e5aq-a4j2	asset per device; filings elevator_periodic/_cat1/_cat5	dob-elevator:<device_number>[:<kind>]	bin, bbl
dob-boilers	DOB NOW: Safety Boiler	52dp-yji6	asset boiler; filing boiler_inspection	dob-boiler:<boiler_id> (asset) / dob-boiler:<boiler_id>:inspection (latest report)	bin_number only
dob-facades	DOB NOW: Safety – Facades Compliance Filings	xubg-57si	filing fisp	dob-fisp:<bin> (latest report)	bin; BBL from borough/block/lot
ll84-benchmarking	LL84 Energy and Water Data Disclosure, 2023 to present	5zyy-y8am	filing ll84_benchmarking per building	ll84:<property_id> (latest year)	nyc_building_identification and nyc_borough_block_and_lot, both ; lists
dohmh-cooling-towers	NYC Cooling Tower Registrations	y4fw-iqfr	asset cooling_tower	dohmh-cooling-tower:<system_id>	bin, bbl
dohmh-water-tanks	Self-Reported Drinking Water Tank Inspection Results	gjm4-k24g	asset water_tank; filing water_tank_inspection	dohmh-water-tank:<bin>:<tank_num> (asset) / …:report (latest year)	bin, bbl
dob-now-permits	DOB NOW: Build – Approved Permits	rbx6-tga4	filing work_permit	dob-now-permit:<work_permit>	bin; bbl, else from borough/block/lot
dob-bis-permits	DOB Permit Issuance	ipu4-2q9a	filing work_permit	dob-bis-permit:<job__>:<job_doc___>:<permit_type>:<permit_subtype>	bin__; bbl, else from borough/block/lot
dob-certificates-of-occupancy	DOB certificates of occupancy (BIS and DOB NOW)	bs8b-p36w + pkdm-hqz6	filing certificate_of_occupancy	dob-co:<job>	bin_number (bs8b-p36w) or bin; bbl, else from borough/block/lot
dob-elevator-details	DOB NOW elevator device details	juyv-2jek	filing elevator_device_details, grouped by device_id	dob-elevator-details:<device_id>	none: the device, as dob-elevators loaded it
dohmh-cooling-tower-inspections	Cooling tower inspections	f9wb-g8mb	filing cooling_tower_inspection, grouped by system_id, one per inspection	dohmh-cooling-tower-inspection:<system_id>:<day>	bin, bbl
dohmh-water-tank-results	Drinking water tank inspection results	rytv-g5ui	filing water_tank_compliance, grouped by bin	dohmh-water-tank-result:<bin>:<year>:<activity>	bin, bbl
dep-boiler-permits	DEP boiler permits (CATS)	f4rp-2kvy	filing dep_boiler_permit	dep-cats:<applicationid>	bin; BBL from borough/block/lot
hpd-jurisdiction	Buildings Subject to HPD Jurisdiction	kj4p-ruqc	designation hpd_jurisdiction	hpd-jurisdiction:<buildingid>	bin; BBL from boroid/block/lot
dof-exemptions	DOF property exemptions	muvi-b6kx	filing tax_exemption, grouped by lot (parid) or condo	dof-exemption:<lot or condo>:<code>	BBL only; condo units via the billing lot in p8u6-a6it
dof-abatements	DOF property abatements	rgyu-ii48	filing tax_abatement, grouped by lot (parid) or condo	dof-abatement:<lot or condo>:<code>	BBL only; condo units via the billing lot in p8u6-a6it
fdny-mandatory-inspections	FDNY mandatory inspections	kfgh-h6re	filing fdny_inspection	fdny-mandatory-inspection:<insptn_ops_detail>	bldg_current_bin, bbl
fdny-risk-inspections	FDNY risk-based inspections (RBIS)	itd7-gx3g	filing fdny_inspection	fdny-risk-inspection:<insptn_ops_detail>	bldg_current_bin_fk, bbl
fdny-building-summary	FDNY Bureau of Fire Prevention building summary (historical, 2019)	nvgj-hbht	asset sprinkler, asset standpipe	fdny-building-summary:<bin>:<system>	bin, bbl

Quirks found sampling these (2026-10-08): a boiler inspection_date of 01/01/1900 means not yet inspected; new-boiler registrations (report_type BNR) have no boiler_id; the bedbug dataset repeats some rows verbatim; DOB uses 1000000 to 5000000 as "BIN unknown".

HPD violations, wvxf-dwi5

Housing Maintenance Code violations, including the lead orders. Filter by bin; page by violationid; backfill window on novissueddate.

Columns: violationid, buildingid, registrationid, boroid, boro, housenumber, lowhousenumber, highhousenumber, streetname, streetcode, zip, apartment, story, block, lot, class, inspectiondate, approveddate, originalcertifybydate, originalcorrectbydate, newcertifybydate, newcorrectbydate, certifieddate, ordernumber, novid, novdescription, novissueddate, currentstatusid, currentstatus, currentstatusdate, novtype, violationstatus, rentimpairing, latitude, longitude, communityboard, councildistrict, censustract, bin, bbl, nta.

Field	Meaning and mapping
violationid	Unique id; source_id.
bin, boroid + block + lot	Resolution keys: BIN, then a BBL composed as borough + 5-digit block + 4-digit lot.
class	Severity A, B, C, or I. Used as citation_code when there is no order number.
ordernumber	HPD order code; citation_code. Lead orders: 616 presumed lead, 617 lead hazard, 618 RPO filing (child at risk), 619 RPO annual certification, 620 RPO turnover certification, 621 turnover abatement, 622 turnover certification failure, 623 turnover clearance failure, 624 inconclusive XRF, 625 turnover notice failure, 626 child blood lead. These become event_type; any other order is violation.
novdescription	Notice text; description.
approveddate, novissueddate	issued_date (approved date preferred).
newcorrectbydate, originalcorrectbydate	due_date: the new date if present, else the original.
certifieddate	Set when the owner certified correction; used for overdue logic and as closed_date when not open.
currentstatusid, currentstatus, violationstatus	Status. The id is authoritative (see below); the text and violationstatus are fallbacks.
apartment	Unit label; linked to units at persist time, subject to the creatable-label guard.

Status ids map to status_detail: 1 OPEN; 2 OPEN_NOV_SENT; 3 OPEN_CERTIFIED_ON_TIME; 4 OPEN_CERTIFIED_LATE; 6 OPEN_POSTPONEMENT_GRANTED; 7 OPEN_POSTPONEMENT_DENIED; 8 OPEN_FALSE_CERT; 10 OPEN_CIV14_MAILED; 11 OPEN_WILL_BE_REINSPECTED; 20 OPEN_DEFECT_LETTER; 21 OPEN_NOT_COMPLIED; 22 and 23 OPEN_NO_ACCESS_1ST / _2ND; 24 OPEN_REOPENED; 27 OPEN_INFO_NOV_SENT; 28 OPEN_INVALID_CERT; 36 OPEN_PARTIAL_ACCESS (all OPEN); 19 is CLOSED; 9 is DISMISSED. Nothing about the deadline is stamped into status_detail: whether a row is overdue is computed when read, by one rule (domain/src/violations/deadlineRule.ts). Rows synced before October 2026 may still end in _OVERDUE until they are synced again; every reader ignores that suffix. Unrecognised ids fall back to violationstatus, then keyword matching on currentstatus, then plain OPEN.

HPD complaints and problems, ygpa-z7cr

One row per complaint problem. Page by complaint_id; backfill window on received_date (the one date every row carries; complaint_status_date is null while open). Rows carry bin and bbl.

Columns: received_date, problem_id, complaint_id, building_id, borough, house_number, street_name, post_code, block, lot, apartment, community_board, unit_type, space_type, type, major_category, minor_category, problem_code, complaint_status, complaint_status_date, problem_status, problem_status_date, status_description, problem_duplicate_flag, complaint_anonymous_flag, unique_key, latitude, longitude, council_district, census_tract, bin, bbl, nta.

Field	Mapping
complaint_id	source_id.
bin	Resolution key.
type, else major_category	event_type (fallback housing_complaint).
major_category, minor_category	description, joined with " - ".
complaint_status	Status: contains CLOSE, RESOLVED or COMPLETED is CLOSED; OPEN, PENDING or ACTIVE is OPEN; anything else is OPEN with detail UNKNOWN.
received_date, complaint_status_date	issued_date; closed_date when not open.
apartment	Unit label.

HPD litigation, 59kj-x8nc

Housing-court cases against owners. Full pull each run; filterable by bin or bbl.

Columns: litigationid, buildingid, boroid, housenumber, streetname, zip, block, lot, casetype, caseopendate, casestatus, casejudgement, findingofharassment, findingdate, penalty, respondent, latitude, longitude, community_district, council_district, census_tract, bin, bbl, nta.

Mapping: litigationid is source_id; casetype is event_type and citation_code; caseopendate is issued_date; casestatus containing DISMISS is DISMISSED, containing CLOSED is CLOSED, otherwise OPEN; description appends "judgement entered" when casejudgement is YES. Resolves by bin, then bbl.

DOB violations, 3h2n-5cm9

Dates are compact YYYYMMDD text with "0" as no date. Backfill window on issue_date.

Columns: isn_dob_bis_viol, boro, bin, block, lot, issue_date, violation_type_code, violation_number, house_number, street, disposition_date, disposition_comments, device_number, description, ecb_number, number, violation_category, violation_type.

Mapping: source_id uses number; event_type is violation_category; citation_code is violation_type; issue_date is issued_date; disposition_date is closed_date. Status: a dismiss keyword in category is DISMISSED; a disposition date or a "resolved" category is CLOSED; otherwise OPEN. Resolves by bin.

DOB complaints, eabe-havv

Replaced wholesale each night: all 3.1M rows share one :updated_at, :created_at and dobrundate (checked 2026-09-30), so no Socrata field says what changed. Only about 21k rows are ACTIVE; the rest are CLOSED. The nightly delta is status = 'ACTIVE' plus any row whose date_entered, inspection_date or disposition_date falls from 14 days before the watermark through today. All three dates are MM/DD/YYYY text, so the backfill windows on date_entered with LIKE / IN patterns; impossible values such as 20/22/2022 become null dates.

Columns: complaint_number, status, date_entered, house_number, house_street, zip_code, bin, community_board, special_district, complaint_category, unit, disposition_date, disposition_code, inspection_date, dobrundate.

Mapping: complaint_number is source_id; complaint_category is event_type and citation_code; date_entered is issued_date; disposition_date is closed_date. Status from status and disposition date: dismiss keyword is DISMISSED, a disposition date or "clos" / "resolved" is CLOSED, otherwise OPEN. Resolves by bin.

ECB violations, 6bgk-3dad

The money and hearings feed. Compact YYYYMMDD dates with "0" as no date. Backfill window on issue_date.

Columns: isn_dob_bis_extract, ecb_violation_number, ecb_violation_status, dob_violation_number, bin, boro, block, lot, hearing_date, hearing_time, served_date, issue_date, severity, violation_type, respondent_name, respondent_house_number, respondent_street, respondent_city, respondent_zip, violation_description, penality_imposed (sic), amount_paid, balance_due, infraction_code1 to infraction_code10 with section_law_description1 to 10, aggravated_level, hearing_status, certification_status.

Mapping: ecb_violation_number is source_id; violation_type is event_type and citation_code; issue_date is issued_date; hearing_date is due_date; violation_description is description. Status: ecb_violation_status = RESOLVE is CLOSED, or DISMISSED when hearing_status says dismissed or vacated; ACTIVE is OPEN (a passed hearing is not stamped: the app shows it as "hearing passed", computed when read); otherwise hearing_status decides. The map reads the hearing date (due_date) and the promoted agency_status_text (hearing_status) and balance_due columns. Resolves by bin.

OATH hearings, jz4z-kudi

Tickets from all issuing agencies; the feed excludes issuing_agency LIKE '%BUILDINGS%' (DOB tickets arrive through ECB). Backfill window on violation_date, not hearing_date, which is null until a hearing is scheduled. :updated_at is a real per-row stamp, except that the city occasionally republishes the whole dataset under one stamp (2026-10-09 00:25 UTC, 20.8M rows); the delta skips such a stamp.

Columns: ticket_number, violation_date, violation_time, issuing_agency, respondent_last_name, balance_due, violation_location_borough, violation_location_block_no, violation_location_lot_no, violation_location_house, violation_location_street_name, violation_location_floor, violation_location_city, violation_location_zip_code, violation_location_state_name, respondent address columns, hearing_status, hearing_result, scheduled_hearing_location, hearing_date, hearing_time, decision columns, total_violation_amount, violation_details, date_judgment_docketed, penalty_imposed, paid_amount, additional_penalties_or_late_fees, compliance_status, violation_description, and charge_1_* to charge_10_* (code, section, description, infraction amount).

Mapping: ticket_number is source_id; issuing_agency chooses agency per row (FIRE is fdny; SANITATION, RECYCLING or POLICE is dsny; DOHMH, DOH or HEALTH is dohmh; anything else is skipped); event_type is oath_hearing; violation_date is issued_date; hearing_date is due_date. Status: dismissal or vacatur in hearing_result / compliance_status is DISMISSED; "closed", "paid in full", "satisfied", a settled balance, or paid in full is CLOSED, and so is "All Terms Met" unless a balance is still due; otherwise OPEN. Resolution keys are the BBL composed from violation_location_borough (borough names, STATEN IS for Staten Island), violation_location_block_no and _lot_no, then the house number and street plus borough, the lowest-confidence tier. Block 00000 and lot 0000 are blanks and NOT NYC has no borough code, so those rows resolve by address only (#93).

311 service requests, erm2-nwe9

Filtered to agency IN ('DEP','DSNY','FDNY','DOT'). HPD and DOB requests are the same problems ygpa-z7cr and eabe-havv carry as complaints (HPD's unique_key is the 311 number), so they are left out. Resolution keys are bbl, then incident_address plus borough when bbl is blank (#93). The dataset updates several times a day. Backfill window on created_date (default window 7 days, the densest feed).

Columns include: unique_key, created_date, closed_date, agency, agency_name, complaint_type, descriptor, descriptor_2, location_type, incident_zip, incident_address, street_name, cross_street_1, cross_street_2, address_type, city, status, due_date, resolution_description, resolution_action_updated_date, community_board, council_district, police_precinct, bbl, borough, coordinates, latitude, longitude, location, and channel, park, vehicle, taxi and bridge columns that the pipeline does not read.

Mapping: unique_key is source_id; the agency code maps to the lowercase agency id; complaint_type is event_type; created_date and closed_date are issued_date and closed_date; descriptor is description. Status: closed_date set or status "closed" is CLOSED; "cancel" is DISMISSED (detail CANCELLED); otherwise OPEN. Resolution keys are incident_address plus borough.

DOF tax lien sale list, 9rz4-mjek

The lots DOF noticed for a tax lien sale: one row per lot per list (cycle: 90, 60, 30, 10 Day Notice, Final Sale; spelled "Days" in 2020) and month, with no amounts. Sale years so far: 2019, 2020, 2021, 2025; last published 2025-06. A lot can appear up to 19 times.

Columns: month, cycle, borough, block, lot, tax_class_code, building_class, community_board, council_district, house_number, street_name, zip_code, water_debt_only.

Mapping: one event per lot. A 10-digit BBL is composed from borough, block, lot and is both the resolution key and the source_id (dof-lien:<bbl>). The lot's newest list (latest month) gives status_detail and agency_status_text (its cycle); its first list is issued_date. OPEN while that newest list is in the dataset's newest sale year, CLOSED otherwise. event_type is tax_lien, or tax_lien_water_debt when water_debt_only is YES; description names the newest list, the sale years and the building class or address. A row with no BBL is skipped.

Building footprints, 5zhs-2jue

Layer 1 identity source: one row per building.

Columns: the_geom, name, bin, doitt_id, shape_area, base_bbl, objectid, construction_year, feature_code, geom_source, ground_elevation, height_roof, last_edited_date, last_status_type, mappluto_bbl, shape_length.

Field	Use
bin	The building's identity; a row without one is dropped.
mappluto_bbl, else base_bbl	Tax lot for the PLUTO join (10 digits).
construction_year	Year built fallback when PLUTO has none.

PLUTO, 64uk-42ks

Tax-lot facts; 108 columns, of which the pipeline reads bbl, address, borough, zipcode, latitude, longitude, yearbuilt and unitstotal. PLUTO's bbl is a decimal string (for example 4087860042.00000000), truncated and left-padded to 10 digits. PLUTO has no BIN, and one lot can hold several buildings, so a footprint row is joined to its lot's facts on BBL: yearbuilt becomes buildings.year_built and unitstotal the unit count. A footprint with no PLUTO match still becomes a buildings row, without address and year facts.

Field-name gotchas

• penality_imposed in ECB is misspelled in the source.
• DOB and ECB use "0" for no date, and DOB contains impossible dates.
• HPD violation status text is unreliable; use currentstatusid.
• Datasets that expose a BBL do not all format it alike: HPD violations compose it from boroid, block, lot; PLUTO's is decimal; Footprints' is already 10 digits.