electricsheepafrica/african-anti-corruption-enforcement

---
license: cc-by-4.0
task_categories:
- tabular-classification
- tabular-regression
tags:
- governance
- anti-corruption
- accountability
- sub-saharan-africa
- synthetic
- lmic
- transparency
- enforcement
- public-sector
pretty_name: African Anti-Corruption Enforcement
size_categories:
- 10K<n<100K
configs:
- config_name: baseline
data_files: data/baseline.csv
- config_name: strengthened_enforcement
data_files: data/strengthened_enforcement.csv
- config_name: weakened_accountability
data_files: data/weakened_accountability.csv
---

# African Anti-Corruption Enforcement

Synthetic dataset modelling anti-corruption enforcement outcomes across 12 Sub-Saharan African countries under three policy scenarios. Parameters are anchored to real-world statistics from Transparency International, the EFCC (Nigeria), SIU/Zondo Commission (South Africa), EACC (Kenya), the African Union's Common African Position on Asset Recovery, and the G20 2025 Accountability Report on Whistleblower Protection.

## Dataset Summary

| Property | Value |
|---|---|
| Total records | 30,000 (10,000 per scenario) |
| Countries | 12 SSA nations |
| Variables | 15 per record |
| Scenarios | baseline, strengthened_enforcement, weakened_accountability |

## Scenarios

- **baseline** – Calibrated to observed 2023–2025 enforcement statistics. Median prosecution rate ~25%, median asset recovery rate ~7%.
- **strengthened_enforcement** – Simulates higher agency independence (+18), stronger whistleblower protection (+20), increased prosecution (+45%), doubled asset recovery, and +8 CPI points.
- **weakened_accountability** – Models institutional erosion: agency independence −15, whistleblower protection −15, prosecution −40%, asset recovery −60%, and −10 CPI points.

## Countries

Nigeria, South Africa, Kenya, Ghana, Tanzania, Uganda, Ethiopia, Senegal, Mozambique, Zambia, Rwanda, Botswana.

## Variables

| Variable | Type | Description |
|---|---|---|
| `country` | str | Country name |
| `scenario` | str | Policy scenario |
| `corruption_perception_index` | float | TI CPI score (0–100, higher = cleaner) |
| `cases_investigated` | int | Number of corruption cases under investigation |
| `cases_prosecuted` | int | Cases forwarded to prosecution |
| `prosecution_rate` | float | cases_prosecuted / cases_investigated |
| `convictions` | int | Number of convictions secured |
| `conviction_rate` | float | convictions / cases_prosecuted |
| `assets_confiscated_usd_millions` | float | Value of confiscated assets (USD millions) |
| `asset_recovery_rate` | float | Fraction of estimated stolen assets recovered |
| `whistleblower_reports` | int | Number of whistleblower submissions |
| `whistleblower_protection_score` | float | Protection framework quality (0–100) |
| `agency_independence_score` | float | Anti-corruption agency independence (0–100) |
| `enforcement_effectiveness_score` | float | Composite effectiveness index (0–100) |
| `enforcement_class` | str | strong / moderate / weak |

## Research Sources

1. **Transparency International CPI 2025** – SSA average score 32/100; Seychelles (68) highest, Somalia/South Sudan (9) lowest.
2. **EFCC Nigeria 2024** – 15,724 petitions → 12,928 investigated → 5,083 prosecuted → 4,111 convictions; $214.5M recovered.
3. **South Africa SIU 2023/24** – 1,919 investigations closed, 583 criminal referrals, R8B saved, ~80% conviction rate in commercial cases.
4. **Kenya EACC 2023/24** – 5,171 reports, 534 files under probe, 26.7% conviction rate, Ksh2.9B recovered.
5. **AU Common African Position on Asset Recovery** – Africa loses ~$150B annually through illicit financial flows; asset recovery typically <10%.
6. **G20 2025 Accountability Report on Whistleblower Protection** – Most SSA countries lack dedicated whistleblower legislation; protection gaps in developing countries.

## Usage

```python
from datasets import load_dataset

ds = load_dataset("electricsheepafrica/african-anti-corruption-enforcement", "baseline")
df = ds["train"].to_pandas()
```

## Generation & Validation

```bash
pip install -r requirements.txt
python generate_dataset.py --scenario all --n-records 10000
python validate_dataset.py --data-dir data --plot-dir plots
```

## License

CC-BY-4.0