---
license: cc-by-4.0
language:
  - en
pretty_name: Warnely Country Risk Dataset
size_categories:
  - n<1K
task_categories:
  - tabular-classification
  - tabular-regression
tags:
  - travel
  - safety
  - geography
  - country-risk
  - government-advisories
  - global-peace-index
configs:
  - config_name: default
    data_files:
      - split: train
        path: data.csv
dataset_info:
  features:
    - name: iso2
      dtype: string
    - name: iso3
      dtype: string
    - name: name
      dtype: string
    - name: region
      dtype: string
    - name: risk_score
      dtype: int32
    - name: risk_tier
      dtype: string
    - name: editorial_score
      dtype: int32
    - name: authority_score
      dtype: int32
    - name: structural_score
      dtype: int32
    - name: live_score
      dtype: int32
    - name: fcdo_status
      dtype: string
    - name: state_level
      dtype: int32
    - name: drug_law_tier
      dtype: string
    - name: lgbtq_legal
      dtype: string
    - name: lgbtq_social
      dtype: string
    - name: women_safety_tier
      dtype: string
    - name: gpi_rank
      dtype: int32
    - name: last_reviewed
      dtype: string
    - name: next_review_by
      dtype: string
  splits:
    - name: train
      num_examples: 180
---

# Warnely Country Risk Dataset

180 countries scored for travel safety across editorial, government-advisory, structural, and live-incident dimensions. Composite risk score on a 0-100 scale (lower is safer). Published under CC BY 4.0.

## What's in it

One row per country, 19 columns. Composite formula: `0.5 × editorial + 0.3 × authority + 0.15 × structural + 0.05 × live`, renormalised when components are missing.

Coverage:

- Africa: 49
- Europe: 42
- Latin America & the Caribbean: 33
- Asia: 29
- Middle East: 15
- Oceania: 7
- Unknown: 3
- North America: 2

## Columns

| Column | Description |
|---|---|
| `iso2` | ISO 3166-1 alpha-2 country code (e.g. `TH`, `GB`) |
| `iso3` | ISO 3166-1 alpha-3 country code (e.g. `THA`, `GBR`) |
| `name` | Country name (en-GB) |
| `region` | One of: Africa, Asia, Europe, Latin America, Middle East, North America, Oceania |
| `risk_score` | Composite 0-100 (lower is safer) |
| `risk_tier` | One of: Very Safe, Exercise Normal Caution, Exercise Increased Caution, High Risk, Do Not Travel |
| `editorial_score` | Editorial component 0-100 (six-category traveller-risk roll-up) |
| `authority_score` | Authority component 0-100 (worst-of FCDO + US State advisory) |
| `structural_score` | Structural component 0-100 (GPI + WGI blend; blank if both absent) |
| `live_score` | Live component 0-100 (30-day rolling incident pressure) |
| `fcdo_status` | Semicolon-separated UK FCDO advisory codes |
| `state_level` | US State Department advisory level 1-4 |
| `drug_law_tier` | One of: death, severe, strict, moderate, decriminalised, legal |
| `lgbtq_legal` | Six-tier legal status (marriage / partnership / legal / hostile / criminalised / death) |
| `lgbtq_social` | Five-tier social climate (progressive / accepting / mixed / conservative / hostile) |
| `women_safety_tier` | Five-tier women's-safety rating |
| `gpi_rank` | Global Peace Index rank (latest edition) |
| `last_reviewed` | ISO date of the most recent editorial review |
| `next_review_by` | ISO date of the next scheduled review |

## Sources

- **UK FCDO** travel advice (`gov.uk/foreign-travel-advice`), scraped daily.
- **US State Department** travel advisories (`travel.state.gov`), scraped daily.
- **Global Peace Index** (Institute for Economics and Peace), latest annual edition.
- **World Governance Indicators** (World Bank), latest annual edition.
- **Warnely incident wire** (Reuters, BBC, AP, AFP, USGS, GDACS, ReliefWeb), 30-day rolling.

Full methodology with anchor IDs per component: <https://warnely.com/guides/methodology>.

## Update cadence

Authority signals (FCDO + US State) refresh daily. The editorial layer is reviewed on a tier-based cadence (90 / 180 / 365 days). The composite score is recomputed on every refresh.

This dataset card was generated on 2026-05-14 from the 2026-04-24 editorial snapshot. The dataset itself updates automatically when Warnely rebuilds.

## Use

```python
from datasets import load_dataset
ds = load_dataset("warnely/country-risk")
print(ds["train"][0])
```

Or in JavaScript / TypeScript:

```ts
const res = await fetch("https://warnely.com/methodology/data.csv");
const csv = await res.text();
```

## Citation

> Warnely. 2026. *Warnely Country Risk Dataset*. CC BY 4.0. https://warnely.com/guides/methodology

## Licence

Creative Commons Attribution 4.0 International (CC BY 4.0). Attribute as: *Warnely, https://warnely.com/guides/methodology*.

Note: `fcdo_status` is a derived structural code, not the original FCDO copy. The full FCDO and US State Department advisory text remains Crown / US Government work and is linked from each country page; the dataset only carries the structural code Warnely uses for the composite formula.

## Limitations

- The composite score is a starting point, not a verdict. A 25 in France and a 25 in Costa Rica reflect very different threat profiles; the structured columns are designed to be read together rather than reduced to a single number.
- Structural data (GPI, WGI) lacks coverage for around 17 small Pacific island states; those rows have empty `structural_score` and the composite weights are renormalised.
- This data is not a substitute for the official government advisory. Always check the underlying advisory before booking and again before departure.

## Maintainer

Warnely. Corrections welcome: <hello@warnely.com>.