Glossary
Key terms used across Road Risk
Short definitions for modelling, mapping, physics, statistics, evidence, and interpretation terms used in the public site and live app.
Quick glossary navigation
Jump to the terms you need first
Judge-critical terms
Key terms before reading the results
Examines possible comparative signals before collision-history calibration is introduced.
A model output used for comparison under defined assumptions; not an official road-safety rating.
The main reporting scale in the app and Results page. It is not an observed crash rate.
A historical collision record from RSA source data used as observed context, not Road Risk model output.
Sample-dependent context. A high percentile means high within that comparison set, not proven real-world danger.
A summary of completeness and fallback reliance; it is not proof of real-world accuracy.
Road Risk currently has internal consistency testing; full operational validation remains a required next step before operational use.
Sensitive terms
Warnings for terms that are easy to over-read
This is a friction-limited model estimate, not a legal speed limit or driving recommendation.
A high percentile means high within that comparison set, not proven real-world danger.
This is a reporting scale for comparison, not an observed annual crash probability.
This is historical observed context. It does not prove Road Risk model correctness or official road danger.
Term groups
Model, physics, statistics, mapping, and data quality
- Prospective model
- Examines possible comparative signals before collision-history calibration is introduced.
- Comparative output
- A model value used for comparison under defined assumptions; not an official safety rating.
- Annualised Comparative Model Output
- The primary reporting scale in the app and Results page. It is not an observed annual crash probability.
- Daily interpretive view
- A secondary view of the same model output intended to make small values easier to read.
- Curvature
- How strongly a road changes direction over distance.
- Radius of curvature
- The estimated bend radius; smaller radius generally means a tighter curve.
- Safe speed
- A friction-limited curve-speed estimate; not a legal speed limit.
- Stopping distance
- Reaction distance plus braking distance under active assumptions.
- Percentile
- Position inside a sampled comparison set. It is sample-dependent, not a universal ranking.
- Distribution
- The spread of comparative model outputs across sampled roads.
- ECDF / CDF
- Cumulative graph views showing the proportion of samples below or above a selected output.
- OpenStreetMap
- The open map-data source used for public road geometry and tags.
- OSM tag
- A public map-data attribute. Missing tags are unknown context, not evidence that a feature is absent.
- Overpass API
- A public query service used to retrieve OSM road ways, geometry, and tags.
- GeoJSON
- A standard format for exporting geographic geometry and properties.
- CSV
- A tabular export format for reviewable model values and metadata.
- Source Manifest
- Metadata describing where an export came from, how it can be loaded, and what privacy flags apply.
- RSA Recorded Collision
- A historical RSA source record used as observed context, not a Road Risk model output.
- Observed Collision Context
- Nearby RSA records counted within an explicit radius and active filter set.
- Model-to-Observed Comparison
- A careful comparison between model samples and observed historical records; not proof of model validity.
- Evidence Package
- A review bundle containing Road Risk outputs, assumptions, context, and source metadata.
- Data confidence
- A caution label based on geometry quality, missing tags, fallback assumptions, and completeness.
- Fallback assumption
- A visible substitute used when public data is missing or incomplete.
- Calibration
- Adjusting a model against empirical data. Full collision-history calibration remains a required next step before operational use.
- Validation
- Testing whether outputs behave correctly and match external evidence.
Definitions
Additional model, data, and interpretation terms
Roadway Infrastructure Safety Kinematics, the SciFest research project behind Road Risk.
A comparative indicator under defined assumptions, not an observed crash rate or official road-safety rating.
A secondary conversion of the annualised model output to make small values easier to read.
A measure of how strongly a road changes direction over distance.
The estimated radius of a bend; smaller radius usually means a tighter curve.
The change in road direction across the analysed segment.
The lateral acceleration needed to follow a curve at a given speed.
The adjusted friction value after surface, weather, or scenario effects are considered.
A friction-limited curve-speed estimate; not a legal speed limit.
Reaction distance plus braking distance under the active speed, friction, and selected assumptions.
A controlled adjustment for rain, fog, fatigue, overspeed, BAC, vehicle type, lighting, surface, or related assumptions.
An estimated exposure or context value used when direct traffic counts are unavailable.
The spread of comparative model outputs across sampled roads.
A reachability shape showing areas accessible within a time or distance condition.
A sampled local road-context layer around a chosen place. It is a modelled context view, not an official hotspot declaration.
A modelled section of a generated route used for route comparison under the active assumptions.
A historical recorded collision record from RSA source data. It is observed context, not a model output.
A summary of RSA recorded collisions near a selected road or sample geometry under an explicit search radius.
A comparison between modelled samples and nearby historical collision records. It supports review and calibration, not proof of causation.
The record metadata that explains export kind, source app, schema/model version, privacy flags, hashes, and loadability.
A structured Road Risk export that bundles model output, assumptions, graphs, Maths traces, and context for review.
The open map-data source used for public road geometry and tags.
A public map-data attribute such as speed, surface, lighting, or road class. Missing tags are unknown context, not proof of real-world absence.
A public query service used to retrieve OSM road ways, geometry, and tags.
A standard format for exporting geographic geometry and properties.
Adjusting a model against empirical data. Full collision-history calibration has not yet been completed.
Testing whether outputs behave correctly and match external evidence. Road Risk currently has internal consistency testing, not operational validation.