Glossary
Block-level geocoding
Block-level geocoding publishes each incident with coordinates rounded to a city block - LAPD pre-rounds addresses to the hundred block and data.police.uk snaps points to streets - so no exact address survives in the open data. Analysis therefore works at block, LSOA or neighbourhood scale rather than premises level.
What is block-level geocoding?
Block-level geocoding is the practice of publishing a geographic event with its position deliberately degraded to a small area rather than a point address. Two cataloged releases document it precisely.
The LA Crime Data (2020 to Present) record describes coverage as "City of Los Angeles (LAPD jurisdiction), geocoded to block level with lat/lon", and its terms note that "addresses are pre-rounded to the hundred block and locations anonymized for privacy". Across 1,004,894 incidents, every coordinate sits somewhere on its hundred block.
data.police.uk takes the same approach differently: granularity is "individual incidents with anonymised point locations snapped to streets/LSOAs", across 44 forces and 2021 LSOA boundaries, monthly from July 2023 to June 2026 as published at research time.
Why does block-level geocoding matter when choosing a dataset?
It sets the ceiling on spatial resolution before you buy anything. Teams that plan premises-level routing or alarm-response territory from these feeds discover the limit only after building pipelines.
- Premises analysis is out of scope. A coordinate rounded to the hundred block cannot tell you which building on the block an incident occurred at.
- Zone choice becomes a design decision. Counts per block, LSOA or neighbourhood are the honest aggregation units; anything finer manufactures false precision.
- Privacy compliance comes built in. Because rounding happens upstream, redistribution under commercial delivery terms (Los Angeles) or OGL v3.0 (UK) does not reintroduce personal-location risk.
Both sources score well - 8/10 and 9/10 on Datadory's rubric - but their fitness is for area-level demand modeling, not dispatch mapping.
How do you evaluate block-level geocoding in a data source?
- Read the rounding rule in writing. LAPD's record states the hundred-block rule outright; ask any vendor to state theirs as plainly.
- Check the coordinate spread empirically. Pull a sample via Socrata SoQL ($select, $where, $limit) and confirm points cluster on intersections rather than mid-parcel positions.
- Match your zones to the publication geography. The UK release publishes against 2021 LSOAs - aggregating to older boundaries introduces slivers.
- Verify freshness separately from precision. The LA table is frozen after LAPD's NIBRS transition (last rows 2024-12-30); pair it with any successor NIBRS feed for current months.
- Confirm the license survives redistribution. commercial delivery terms 1.0 and Open Government Licence v3.0 both permit commercial reuse with attribution where required.
Where to see it in context: security-alarm-services data.
Related terms
Entries adjacent to block-level geocoding:
- street-level-crime-data — the incident-level product behind all 1,004,894 LA rows
- lsoa-geography — the 2021 UK zones snapped points resolve to
- socrata-open-data-api — the API layer serving the LA release
Frequently asked questions
Can I recover an exact address from block-level crime data?
No. The Los Angeles release states addresses are pre-rounded to the hundred block and locations anonymized for privacy, and data.police.uk snaps point locations to streets and LSOAs. The precision is destroyed before publication, not merely hidden.
How many incidents does the LA dataset cover?
LA Crime Data (2020 to Present) carries 1,004,894 records across 28 columns, spanning 2020-01-01 to 2024-12-30, roughly 1 GB as CSV, published under commercial delivery terms with full Socrata API access.
What geography should I aggregate block-level points to?
Anything at or above the rounding unit: blocks in Los Angeles, streets or 2021 LSOAs in the UK release. Premises-level claims are not supportable from either source, so build demand models on counts per zone instead.
Datasets containing this field
Datasets containing Block-level geocoding
6 datasets carry block-level geocoding in the catalog. Open one, count the fields, judge for yourself.
Florida Alarm Systems Contractors Licensing
IRS Statistics of Income - Nonfarm Sole Proprietorships
LA Crime Data (2020 to Present)
National Crime Victimization Survey (NCVS)
Every listing shows the field dictionary, sample rows, and coverage before you commit. API, files, or your warehouse. Daily, weekly, or hourly.
Get sample rows