Open Road Risk
  • Home
  • Start Here
  • Project
    • Project overview
    • Current model status
    • AI-assisted development
  • Literature
    • Literature overview
    • Literature evidence register
    • AI-assisted literature review
    • Literature-pipeline alignment
    • Crash frequency models
    • Exposure and traffic volume
    • Spatial methods and network risk
    • Junctions and conflict structure
    • Severity modelling
    • Validation and metrics
    • Transferability and open data limits
  • Data Sources
    • Overview
    • STATS19 Collisions
    • OS Open Roads
    • AADF Traffic Counts
    • WebTRIS Sensors
    • Network Model GDB
    • OS Terrain 50 (grade)
    • Deprivation (IoD 2025)
  • Methodology
    • Methodology Overview
    • Joining the Datasets
    • Feature Engineering
    • Empirical Bayes Shrinkage
  • Exploratory Data Analysis
    • Collision EDA
    • Collision-Exposure Behaviour
    • Vehicle Mix Analysis
    • Road Curvature
    • Months and Days of Week
    • Traffic Volume EDA
    • OSM Coverage
  • Models
    • Modelling Approach
    • Stage 1a: Traffic Volume
    • Stage 1b: Time-Zone Profiles
    • Stage 2: Collision Risk Model
    • Facility Family Split
    • Model Inventory
  • Investigations
    • Investigations overview
    • KSI atlas diagnostic
    • Staffordshire data quality
    • Temporal descriptors evaluation
    • AADF counted-only filter
    • Rank stability harness
    • Zero-calibration diagnostic
  • Outputs
    • Key figures
    • Top-risk map
    • QGIS GeoPackage (Kaggle)
  • Tools
    • ukgeo — UK Geocoder
  • Future Work

On this page

  • Interactive Map
  • Summary Tables
  • Data Notes

Top-Risk Road Segments Map

Interactive screening map for the top 1% highest-risk road links identified by the current Open Road Risk model output.
Modified

July 5, 2026

Last updated: 2026-07-05 · Full GB output rebuild: 2026-07-05.

This page maps the global top 1% highest-risk road links after controlling for traffic exposure. The current GB production export uses the XGBoost risk_percentile ranking from data/models/risk_scores.parquet.

These links are screening candidates for inspection, mapping, portfolio review, and demo use. They are not causal proof, crash prediction guarantees, or engineering-audit results. Sparse collision histories should be interpreted cautiously.

The refreshed GB outputs and top-risk web assets were regenerated from the current production score file on 2026-07-05. Short segments under 50 m are present in the full dataset and map layer; they are flagged for manual review before use as case study examples.

The map colours links by log_predicted_score = log1p(predicted_xgb) rather than by risk_percentile, because the global top-1% percentile values are tightly compressed at the top end. The percentile rank remains available in each popup.

Interactive Map

Loading static GeoJSON…

The map uses client-side multi-select filters over one static GeoJSON file. Values within a filter are combined with OR logic; different filters are combined with AND logic. No selection in a filter means all values for that filter.

Summary Tables

Count by family

Count by road classification

Count by road archetype

Numeric summary

Data Notes

  • Source layer: data/outputs/top_1pct_risk_segments.parquet.
  • Web layer: data/outputs/web/top_1pct_risk_segments.geojson.
  • Context network layers: data/outputs/web/context/context_manifest.json plus split GeoJSON files. Motorway and trunk-A links are shown in full; other_urban and other_rural show the top context percentile by the ranking field recorded in the manifest.
  • Features: 39,413 road links.
  • GeoJSON size: 28,482,475 bytes.
  • Geometry simplification: 25 m tolerance in EPSG:3857, then written as EPSG:4326 with coordinates rounded to 0.00001 degrees.
  • Colour scaling: log_predicted_score = log1p(predicted_xgb).
  • QA status: full-GB production scores and top-risk web assets refreshed on 2026-07-05; top-risk plausibility review still required before publication claims.
  • Background map: OpenStreetMap tiles, with static project GeoJSON files for the risk and context layers.

Open Road Risk

 

Built with Quarto