Loading approved evidence
Retrieving and verifying the requested analytical release. Result data loads before optional map boundaries and source checks.
Your filters and release stay in the address. Navigation remains available.
Retrieving and verifying the requested analytical release. Result data loads before optional map boundaries and source checks.
Your filters and release stay in the address. Navigation remains available.
SOCIAL FACTORS · PHASE 8 · LIVE OPERATIONS
Follow source availability, published-file integrity and the remaining requirements for a stronger study.
Monitoring dates describe operational checks. The published findings remain frozen.
Connected to the existing Lens Master Pipeline. The hourly wake-up checks due sources; new sources use weekly or 30-day intervals. Existing shared-source settings and pauses are preserved. Monthly checks verify six published artifacts. They do not refit a model.
Status read 22 Oct 2026, 22:53 SAST · Live status JSON
A changed catalogue is a prompt to inspect analytical files. It does not establish a new data year or a revised collision rate. A first capture establishes the monitor’s baseline.
| Source and purpose | Status | Last check / success | Next check / interval |
|---|---|---|---|
| DfT road safety tablesNew collision years and revisions; outcomes through 2025 are already inspected. | Change detected | 16 Oct 2026, 08:10 SASTSuccess: 16 Oct 2026, 08:10 SAST | 23 Oct 2026, 08:10 SAST168 hours |
| DfT road traffic downloadsMatched mileage, boundary changes and historical road context. | Change detected | 16 Oct 2026, 08:10 SASTSuccess: 16 Oct 2026, 08:10 SAST | 23 Oct 2026, 08:10 SAST168 hours |
| English indices of deprivationNew domain vintages require timing and geography reconciliation; never backfill future information. | Baseline captured | 10 Oct 2026, 16:04 SASTSuccess: 10 Oct 2026, 16:04 SAST | 9 Nov 2026, 16:04 SAST720 hours |
| Nomis 2011 qualificationsCorrections to the frozen 2011 area descriptor. | Baseline captured | 10 Oct 2026, 16:04 SASTSuccess: 10 Oct 2026, 16:04 SAST | 9 Nov 2026, 16:04 SAST720 hours |
| Nomis Census 2021 bulk dataPotential new education/context inputs require an amended protocol and matched geography. | Baseline captured | 10 Oct 2026, 16:04 SASTSuccess: 10 Oct 2026, 16:04 SAST | 9 Nov 2026, 16:04 SAST720 hours |
| Home Office recorded crime tablesRevised force-level burglary files, recording breaks and new years. | Checked | 21 Oct 2026, 14:08 SASTSuccess: 21 Oct 2026, 14:08 SAST | 28 Oct 2026, 14:08 SAST168 hours |
| Police.uk data changelogForce-month gaps, recording-system changes and crime completeness. | Checked | 17 Oct 2026, 16:13 SASTSuccess: 17 Oct 2026, 16:13 SAST | 24 Oct 2026, 16:13 SAST |
Six published protocols/results/readings are checked monthly against frozen byte hashes. This verifies delivery integrity, not a numerical refit or the current publisher files. Failed attempts retry after 24 hours; every attempt is retained. Source retrieval failures use the existing pipeline backoff.
Unresolved
Joint road-class and urban/rural mileage controls aligned to collision geography and year.
Not yet reserved
A new protocol must freeze outcomes, feature timing and acceptance criteria before inspecting any new holdout. 2023–2025 cannot be reused as unseen data.
Unresolved
Reconcile exact forecast-origin input vintages, revised histories and geographical changes. New deprivation editions are not available retrospectively.
Unresolved
Verify offence definitions, complete force-month reporting, annual population denominators and exact geographic matches. Burglary results do not establish an all-crime effect.
No claims validation recorded
Geocoded insurer exposure and developed claims, fresh assessment, calibration and intended-use review. Public collision data cannot close this gap.
These are evidence requirements, not failures that can be cleared by a successful page check. Fresh validation must be designed before the next outcomes are inspected.
The separate decile study now includes nine geographic assessments, 999 fully refitted regional samples, sensitivity tests and a shared finding register. No decile hypothesis passes every registered requirement.
Phase 5 is complete. Phase 6 has a tested exposure pipeline and a failed real-data gate; LSOA frequency remains unavailable. Phase 7 publishes the authority findings. These decile artifacts are not yet part of the original six-file monthly integrity monitor; that extension is Phase 8.
Explore the decile study and its actual results →
New: adaptive adjacent grouping and predictive comparison →. This separate v3 release also remains outside the original monthly integrity monitor.
Archived initial proposalSource and monthly tasks are deduplicated and stored in Admin → Review → Research tasks. An owner can record a blocker, request a new candidate or record no model change after reviewing the current source. This does not certify a hypothesis or approve a pricing factor.