electricsheepafrica/african-property-insurance-risk

---
license: cc-by-4.0
task_categories:
- tabular-classification
tags:
- insurance
- property
- risk
- flood
- fire
- storm
- sub-saharan-africa
- synthetic
- catastrophe
pretty_name: African Property Insurance Risk
size_categories:
- 10K<n<100K
language:
- en
configs:
- config_name: baseline
data_files: data/property_risk_baseline.csv
- config_name: high_risk_period
data_files: data/property_risk_high_risk_period.csv
- config_name: climate_adapted
data_files: data/property_risk_climate_adapted.csv
---

# African Property Insurance Risk

Synthetic property insurance risk dataset covering 12 Sub-Saharan African countries across 3 underwriting scenarios. Designed for catastrophe risk modeling, premium pricing, loss forecasting, and insurance portfolio management research.

## Dataset Description

- **100,000** synthetic property insurance risk records
- **12 countries**: South Africa, Nigeria, Kenya, Ghana, Tanzania, Uganda, Rwanda, Ethiopia, Senegal, Cote d'Ivoire, Zambia, Mozambique
- **3 scenarios**: baseline, high_risk_period, climate_adapted
- **24 variables** per record

## Variables

| Variable | Type | Description |
|---|---|---|
| policy_id | string | Unique policy identifier |
| country | string | Country of origin |
| region | string | Sub-region (East, West, Southern) |
| year | int | Policy year |
| property_type | string | Property category (residential, commercial, industrial, agricultural) |
| property_value_usd | float | Property insured value in USD |
| construction_type | string | Construction (brick, concrete, wood, metal, mudbrick, mixed) |
| building_age | int | Building age in years (1-80) |
| floors | int | Number of floors (1-20) |
| flood_risk_score | float | Flood risk score (0-1) |
| fire_risk_score | float | Fire risk score (0-1) |
| storm_risk_score | float | Storm/hail risk score (0-1) |
| earthquake_risk_score | float | Seismic risk score (0-1) |
| crime_risk_score | float | Theft/vandalism risk score (0-1) |
| combined_risk_score | float | Weighted aggregate risk score (0-1) |
| premium_usd | float | Annual premium in USD |
| deductible_usd | float | Policy deductible in USD |
| coverage_type | string | coverage level (basic, standard, comprehensive) |
| flood_zone | string | FEMA-style flood zone (A, B, C, X, V) |
| proximity_to_water_km | float | Distance to nearest water body in km |
| elevation_m | float | Property elevation in meters |
| previous_claims_count | int | Number of prior claims (0-10) |
| previous_losses_usd | float | Total prior losses in USD |
| risk_class | string | low_risk, medium_risk, high_risk, extreme_risk |
| scenario | string | baseline, high_risk_period, or climate_adapted |

## Scenarios

- **baseline** (50K records): Standard market conditions, 2020-2024. Risk distribution reflects current underwriting practices.
- **high_risk_period** (30K records): Elevated risk due to climate change impacts, increased extreme weather events, 2023-2025.
- **climate_adapted** (20K records): Climate-adjusted risk assessment with updated models, 2024-2025.

## Generation Methodology

Parameters are calibrated against published data from:
- Santam Insurance Barometer 2024/2025
- Deloitte Africa Insurance Outlook 2024/2025
- South African Insurance Association catastrophe data
- Kenya Insurance Regulatory Authority risk assessments
- Munich Re Africa catastrophe loss data
- World Bank disaster risk financing for Sub-Saharan Africa
- FEMA flood zone classification adapted for African contexts

Risk scores follow beta distributions modulated by property characteristics, geographic factors, and climate projections. Premiums are calculated using actuarial models adjusted for country-specific loss experience.

## Use Cases

- Catastrophe risk modeling and accumulation management
- Climate change impact on property insurance
- Premium pricing optimization
- Cross-country risk comparison
- Portfolio risk segmentation
- Synthetic data generation evaluation

## Citation

```bibtex
@dataset{african_property_risk_2026,
title={African Property Insurance Risk Dataset},
author={ElectricSheepAfrica},
year={2026},
license={cc-by-4.0}
}
```

## License

CC BY 4.0 - This is synthetic data generated for research and educational purposes.