Part of a Data Solution

Postal Codes Solution
Every worldwide postal-code dataset in one bundle — the ultimate global reference for precise postal codes.
Explore →

Premium

You're viewing a free sample

Get the complete dataset — all rows and files — delivered instantly after checkout, with lifetime access to the latest version.

Secure checkout via Stripe
Instant download after payment
Lifetime access to the latest version
Singapore Open Data Licence version 1.0 license

20% off

$99.90$79.92

one-time payment

Download sample · CSV · 9.7 kB Report an Issue

API Access

Access dataset files directly from scripts, code, or AI agents.

Browse dataset files

Dataset Files

Each file has a stable URL (r-link) that you can use directly in scripts, apps, or AI agents. These URLs are permanent and safe to hardcode.

/logistics/postal-codes-sg/

├ .licensing-notes.md

https://datahub.io/logistics/postal-codes-sg/_r/-/.licensing-notes.md

├ .migration-notes.md

https://datahub.io/logistics/postal-codes-sg/_r/-/.migration-notes.md

├ ATTRIBUTION.txt

https://datahub.io/logistics/postal-codes-sg/_r/-/ATTRIBUTION.txt

├ README.md

https://datahub.io/logistics/postal-codes-sg/_r/-/README.md

└ datapackage.yaml

https://datahub.io/logistics/postal-codes-sg/_r/-/datapackage.yaml

Key Files

Start with these files — they give you everything you need to understand and access the dataset.

datapackage.yaml— metadata & schema

https://datahub.io/logistics/postal-codes-sg/_r/-/datapackage.yaml

README.md— documentation

https://datahub.io/logistics/postal-codes-sg/_r/-/README.md

Typical Usage

1. Fetch datapackage.yaml to inspect schema and resources
2. Download data resources listed in datapackage.yaml
3. Read README.md for full context

Data Files

postal-codes-sg-sample

Field	Type	Description	Constraints	Title
country_code	string	ISO 3166-1 alpha-2 country code — always SG	{ "pattern": "^SG$", "required": true }	Country Code
postal_code	string	Six-digit postal code — primary key, one row per code	{ "pattern": "^\\d{6}$", "required": true }	Postal Code
place_name	string	Building name where known, else the road name, else the subzone		Place Name
admin_name1	string	CDC District name (e.g. Central Singapore)		Administrative Name 1
admin_code1	string	CDC District two-digit code (01-05)		Administrative Code 1
admin_name2	string	URA Planning Area name (e.g. Bishan)		Administrative Name 2
admin_code2	string	Reserved — no official Planning Area code system found		Administrative Code 2
admin_name3	string	URA Subzone name — finer than Planning Area (e.g. Bishan East)		Administrative Name 3
admin_code3	string	Reserved — no official Subzone code system found		Administrative Code 3
latitude	number	Latitude (WGS84, decimal degrees) where resolved — see accuracy		Latitude
longitude	number	Longitude (WGS84, decimal degrees) where resolved — see accuracy		Longitude
accuracy	string	BUILDING — per-building point from OneMap; blank if no coordinate was found		Accuracy
alternative_city_name	string	Not available for this dataset — always blank		Alternative City Name

Download

Download sample CSV

About

Last updated: 3 October 2026
Total rows: 100
Format: CSV
File size: 9.7 kB
Sources: OneMap, Singapore Land Authority — postal codes, building and road names, and coordinates; Master Plan 2019 Subzone Boundary (No Sea), Urban Redevelopment Authority, via data.gov.sg — CDC district, planning area and subzone
License: Singapore Open Data Licence version 1.0

About this dataset

Singapore (SG): migration notes

Internal record of the P5 migration from the old scripts to the producer: the discrepancies found, their causes, and who decided what. This is a dated record, so the numbers are as of the migration and not kept current. It is not shipped in the zip.

Issue: pc-28e.7.51 (closed 2026-10-03)
Commit: d62fae6 (worktree commit 55d0071, landed unchanged), batch 4
Parity verdict: format-only (CRLF → LF); passes the gate
Producer: custom, refresh disabled

Sources

Input	URL	Notes
OneMap (Singapore Land Authority) address dump, June 2020 snapshot	https://raw.githubusercontent.com/isen-ng/singapore-postal-codes-1/master/database.json.gz	Static file in a GitHub mirror (141,848 records). No OneMap API call, no crawl, no auth. Postal code, building, road name, coordinates
URA Master Plan 2019 Subzone Boundary (No Sea) GeoJSON	https://api-open.data.gov.sg/v1/public/api/datasets/d_8594ae9ff96d0c708bc2af633048edfb/poll-download	data.gov.sg returns a signed S3 link, which the producer follows; the stable poll URL is recorded in `sg.inputs.json`

Both are under the Singapore Open Data Licence v1.0 (.licensing-notes.md).

The old pipeline didn't fetch anything. build_base.py read a staged export, sources/singapore_fedex_base.csv, produced by an external scraper (fedex-agents-prototype/countries/SGP/scripts/scrape_postal.py), and join_coordinates.py added coordinates from the OneMap dump. The scraper is no longer on this machine.
This is a plan D1 case (a dataset built from a missing intermediate export); the plan names the sg OneMap scrape explicitly. The agent rebuilt the scraper's steps from the two inputs above and checked them against the live zip cell for cell.

Parity against the live zip (2026-10-03)

Rows: 121,513 live and 121,513 fresh, none only on one side. 0 changed cells. With CRs stripped, the fresh CSV is byte-identical to the live one, row order included.

The public sample (postal.datahub.io/sg/sg.csv) came out unexplained (informational only); it wasn't investigated.

Decisions

Decision	By	Basis
Reconstruct the missing scraper step from the original sources rather than hold sg back	Plan decision D1; orchestrator instruction to stop and report if OneMap needed auth or a per-address crawl	Neither was needed: the old pipeline only ever read the static June 2020 dump
Land without asking	Standing rule (row_order/CRLF-only, user, 2026-10-02)	`format-only`, equal rows, 0 changed cells
Keep `admin_code2/3` empty, although the GeoJSON carries URA codes (`PLN_AREA_C`, e.g. `BM`; `SUBZONE_C`, e.g. `BMSZ12`)	Agent: kept for parity	Follow-up filed by the orchestrator as pc-7gv (P4)
Point-in-polygon in pure Python (even-odd ray casting, holes and multipolygons), no shapely	Agent
Remove the row, "zero duplicates" and per-tier row counts from README	Agent, per the README counts rule	`ATTRIBUTION.txt` untouched

Transforms carried over

Reconstructed from the scraper's documented behaviour plus build_base.py and join_coordinates.py (integrity_checker.py, package.py, publish_r2.py and the scripts README deleted):

For each postal code, the first OneMap record whose point falls inside a subzone polygon sets the place and admin columns. Codes with no such record, the "NIL" postal code and records without coordinates are dropped.
place_name = BUILDING, or ROAD_NAME when BUILDING is "NIL".
admin_name2 / admin_name3 = planning area / subzone, title-cased.
admin_name1 / admin_code1 = CDC district from the subzone's URA region: Central → Central Singapore 01, North-East → North East 02, North → North West 03, East → South East 04, West → South West 05.
Coordinates from the record whose BUILDING or ROAD_NAME equals place_name, rounded to 7 dp, blanked outside the bounding box. accuracy = BUILDING; alternative_city_name empty.
Rows ordered by district code, planning area, subzone, then dump order.
Old asserts are now ProducerErrors: known URA region, named subzones, readable inputs, all 5 CDC districts present. No row-count assert. Primary key postal_code.

Open questions

Resolved (user, 2026-10-05, pc-28e.7.59): keep the June 2020 OneMap snapshot (staleness disclosed). Evaluating the OneMap API is tracked as pc-5kk.1.
URA codes for admin_code2/3: open as pc-7gv.
.licensing-notes.md still describes the deleted scripts and carries counts (121,513 rows, 141,848 records); the README Fields table still has a "Fill" column with percentages. Left for the end-of-epic sweep (pc-hfz).

Provenance

Reconstructed 2026-10-05 from the agent transcript d33e1b7e-5027-498a-a97f-ba914ee2524e/subagents/agent-a7a0b891bb088f0a5.jsonl, the orchestrator session d33e1b7e-… (launch instruction, landing, pc-7gv), the plan's D1 entry (session 068ca403-…), the commit message, the bd close reason and docs/parity/p5.md.