← Back to Tranzia

How Safety Scores Work

A transparent, data-driven look at how we evaluate route safety — so you can travel with confidence.

The Factors Framework

We don't just count crimes. We analyze the **Safety Trinity**: Social (Crime), Physical (Traffic), and Environmental (Light).

🛡️40%

Social Safety

Official crime reports (Theft, Assault). Modeled for statistical density.

🚦20%

Physical Safety

Vision Zero data. "Killer Intersections" and high-injury zones are penalized.

💡15%

Luminosity

Context matters. Night penalties apply, but Well-Lit Streets get a "Bright Mode Bonus."

👟15%

Exposure

Street-level vulnerability. Longer walks at night increase risk.

🗣️Display Only

Feedback

We collect user reports, but they don't affect scores yet.

Real World Examples

Scenario A: The "Bright" Path ☀️

11:00 PM • Main Street

Crime (Moderate)Score 8.0
LightingBright Mode (Bonus)
Time ExecutionReduced Penalty
Final Score8.5 (Safe)

Scenario B: The "Killer Intersection" 🚦

2:00 PM • Busy Junction

Crime (Low)Score 9.5
Traffic SafetyHigh Injury Rate
Vision Zero-3.0 Penalty
Final Score6.5 (Caution)

1. Social Safety (Crime) (40%)

Crime data is the backbone of the score.

Where it comes from

  • London: Police.uk street-level incident reports (monthly updates)
  • NYC: NYPD CompStat and precinct-level datasets

How we process it

We map cities into small hexagonal cells (~174m across) and normalize crime counts using percentiles. Crucially, we also model "Ambient Density" for safe areas to account for unreported quality-of-life issues.

Top 5% crime areas → High danger (0.8–1.0)

Average areas → Medium danger (0.4–0.6)

Bottom 25% areas → Low danger (0.0–0.2)

Global Consistency: Apples to Apples

Every city reports crime differently. London includes "Anti-Social Behavior" (noise, loitering), while NYC focuses on "Major Felonies". If we compared raw numbers directly, NYC would look artificially safer than it really is.

Our Harmonization Promise

We calibrate every city individually. A 9.0 Score in Tokyo means the same thing as a 9.0 Score in New York: "Top-tier safety relative to this environment."

2. Physical Safety (Traffic) (20%)

Traffic accidents are a major urban danger.

  • London: TfL Stats19 data tracks injury collisions
  • NYC: Motor Vehicle Collisions (Vision Zero)

High Injury Zone → -3.0 Penalty

Medium Risk Zone → -1.5 Penalty

3. Luminosity (Time & Light) (15%)

Darkness increases risk, but lights mitigate it.

  • Street lamp density (OSM)
  • Time of day execution

Night (8 PM - 5 AM) → Base Penalty

Well-Lit Area → "Bright Mode" (+bonus)

Dark Area → Double Penalty

4. Walking Exposure (15%)

Walking keeps you at street level, exposed to your surroundings.

Walking danger = (walk minutes) / 33

Maximum = 0.5

Applies at night only

5. User Feedback (Display Only)

We collect user feedback but it does not currently affect scores.

😊 Safe😐 Neutral😟 Unsafe

Once we have sufficient volume (5+ reports per segment), feedback may be introduced at a small weight with abuse controls and time decay.

For now, this helps us validate data quality — not influence scores.

The Formula

Total Danger =

(Crime × 0.40) +

(Traffic × 0.20) +

(Time/Light × 0.15) +

(Walk × 0.15) +

(Feedback × 0.10) — not active yet

Safety Score = 10 − (Total Danger × 10)

What the Scores Mean

🟢

8–10: Low Risk

Generally safe, well-traveled routes

🟡

4–7: Caution

Some risk factors — stay alert

🔴

0–4: Elevated Risk

Consider alternatives or extra precautions

Our Commitment to Transparency

Tranzia is built on:

  • Open data from official sources
  • Clear methodology you can understand
  • Community insights that improve accuracy over time
  • Honest, objective assessments — never fearmongering

This is not a black box. You deserve to know how the score is produced.

What We Don't Do

  • ❌ We do not guarantee safety — conditions change
  • ❌ We do not track your location
  • ❌ We never use demographic or personal data
  • ❌ We do not replace your own judgment

Tranzia is a tool — you remain the final expert on your surroundings.