In 95 London areas checked so far, central wards run twenty times the violent-crime rate of the quietest suburbs
Not yet complete for every London ward. This is a validation draft built from 95 web-indexed area pages, not the requested 679-ward official dataset. Periods vary by page (mostly mid-2025 to mid-2026), bands are quintiles of this 95-area sample only, and some rows lack codes, coordinates or component counts. Treat every band as relative to this prototype, not to London.
Asked:“Research and classify every London neighborhood by its level of violent crime to create a color-coded crime map similar to Hoodmaps Crime Mode. Use the most recent complete 12-month period available from official sources such as the Metropolitan Police, City of London Police, data.police.uk, and the London Datastore. Focus on homicide, assault, robbery, sexual offences, and weapons offences. Use wards, LSOAs, or MSOAs as the underlying geographic units, but map them to recognizable neighborhood names wherever possible. Calculate a normalized violent-crime score primarily using offences per 1,000 residents, while also reporting the absolute number of offences, and classify every area into five relative bands: very low, low, medium, high, and very high crime. Return map-ready data containing the neighborhood name, borough, geographic unit and code, latitude and longitude or boundary identifier, violent-crime count, population, crime rate, normalized score from 0 to 100, crime-level band, dominant crime categories, data period, and exact source URLs. Ensure the result covers the whole of London rather than only listing the highest-crime areas, and highlight central nightlife and tourist districts where crime rates may be inflated by large numbers of non-resident visitors.”
95 named London areas across 24 of the 33 boroughs, each with a violent-crime rate per 1,000 residents from ward or district statistics pages covering roughly a rolling 12 months between late 2024 and mid-2026. Rates span 12.3 to 366.9 offences per 1,000; the sample median is 40.4. No row carried usable coordinates, so the map view below is a borough-grouped heat grid, not a boundary map.
Heat grid of all 95 areas, grouped by borough
Tile colour = quintile band within this 95-area sample. Number = violent offences per 1,000 residents. Hover a tile for detail; hover dims other bands. Partial coverage: 9 boroughs have no extracted areas yet.
very lowlowmediumhighvery high◆ = visitor-inflated rate flagged on source page
What the grid cannot say alone
Westminster as a district posts 366.9 offences per 1,000 — but its source explicitly credits tourist crowds and pickpocketing opportunity, so the resident denominator overstates risk to locals. city-guide.info
All six City of London wards in the sample land in the very-high band (74.8–253.8 per 1,000); tiny resident populations of 1,100–8,700 next to huge daytime footfall inflate every rate. streetscan.co.uk
The quietest extracted area, Village ward in Merton, runs 12.3 per 1,000 — one-thirtieth of Westminster — showing how wide the spread is even in a 95-area sample. streetscan.co.uk
The official MPS ward file on the London Datastore lists monthly counts by crime type with its newest coverage spanning Aug 2024 – Jul 2026, so the newest complete rolling 12 months is Aug 2025 – Jul 2026 — the period a production build should use. data.london.gov.uk
What is needed for production: ingest the latest official MPS ward-level CSV directly, filter Aug 2025 – Jul 2026, join all 679 ward codes to one population vintage (2021 Census ward estimates) and ONS ward centroids/boundaries, append City of London Police consistently, then recompute every rate, score and quintile London-wide. Note the MPS page states sexual offences are not supplied at LSOA level, which makes wards the right unit.
All 95 rows
Search by area or borough. Blank cells mean the source page did not state the value.
Area
Borough
Band
Rate /1,000
Offences
Population
Score
Period
Source
Validation prototype: 95 London area rows extracted from web-indexed ward and neighbourhood statistics pages, covering varying rolling 12-month periods (mostly May 2025 – Apr 2026 and Jun 2025 – May 2026). Rate is violent offences per 1,000 residents as stated or derived from stated count and population; the intended violence definition is homicide + violence against the person + robbery + sexual offences + weapons offences, but category completeness varies by page. Normalized score is 100 × percent-rank within the sample; bands are sample quintiles (NTILE 5), not London-wide quintiles. 9 of 33 boroughs have no rows yet; no coordinates were available, so no boundary map is drawn. Official references: MPS Recorded Crime Geographic Breakdown (London Datastore), data.police.uk, and 2021 ward/LSOA population estimates.