Saksham-443paudel/gender-based-violence-ipv

---
license: cc-by-4.0
task_categories:
- tabular-classification
language:
- en
tags:
- healthcare
- gender-based-violence
- intimate-partner-violence
- gbv
- ipv
- mental-health
- sub-saharan-africa
- lmic
pretty_name: "Gender-Based Violence & IPV (Violence Type, Health Consequences, Help-Seeking)"
size_categories:
- 10K<n<100K
configs:
- config_name: urban_one_stop_centre
data_files: data/gbv_urban_one_stop_centre.csv
- config_name: district_health_facility
data_files: data/gbv_district_health_facility.csv
default: true
- config_name: community_level
data_files: data/gbv_community_level.csv
---

# Gender-Based Violence & Intimate Partner Violence Dataset

## Abstract

This dataset provides **30,000 simulated GBV/IPV records** (10,000 per scenario) of women in sub-Saharan Africa. Each record contains 45+ variables including violence type (physical, sexual, emotional, economic), risk factors, injuries, mental health consequences, help-seeking behaviour, barriers, and clinical response. Three settings: urban one-stop centre (26% help-seeking), district facility (20%), and community level (10%).

## 1. Introduction

Approximately 36% of women in SSA experience physical and/or sexual IPV in their lifetime — the highest prevalence globally (WHO 2024). DHS data across 26 SSA countries show physical violence 20-40%, sexual 5-20%, emotional 15-35%. IPV is associated with HIV, depression, PTSD, and adverse pregnancy outcomes. Help-seeking rates remain low due to stigma, fear, normalisation, and lack of services.

**This dataset is entirely simulated. It must not be used for clinical decision-making or legal purposes.**

## 2. Methodology

### 2.1 Parameterization

| Parameter | Value | Source |
| --- | --- | --- |
| IPV lifetime prevalence (SSA) | ~36% | WHO 2024 |
| Physical violence | 20-40% | Frontiers 2023 |
| Sexual violence | 5-20% | Frontiers 2023 |
| Depression among IPV survivors | ~40% | PMC 2020 |
| PTSD among IPV survivors | ~25% | PMC 2020 |
| Help-seeking rate | ~20% | DHS 2019-2024 |
| Partner alcohol as risk factor | OR 1.5-2.0 | PMC 2024 |

### 2.2 Scenario Design

| Scenario | PEP | Counselling | Forensic | Help-Seeking |
| --- | --- | --- | --- | --- |
| Urban one-stop centre | Yes | Yes | Yes | 26% |
| District health facility | Yes | No | No | 20% |
| Community level | No | No | No | 10% |

## 3. Schema

| Column | Type | Description |
| --- | --- | --- |
| id | int | Unique identifier |
| age_years | int | Age |
| marital_status | categorical | married / cohabiting / divorced / single |
| education | categorical | none / primary / secondary / tertiary |
| physical_violence | binary | Physical IPV |
| sexual_violence | binary | Sexual IPV |
| emotional_violence | binary | Emotional IPV |
| economic_violence | binary | Economic violence |
| any_ipv | binary | Any IPV |
| violence_frequency | categorical | once / few_times / frequent |
| perpetrator | categorical | current_partner / former / family / other |
| partner_alcohol_use | binary | Partner alcohol use |
| witnessed_violence_childhood | binary | Childhood violence exposure |
| injury_sustained | binary | Injury from violence |
| injury_type | categorical | bruises / fracture / head_injury / burns / strangulation |
| depression | binary | Depression |
| ptsd | binary | PTSD |
| anxiety | binary | Anxiety |
| suicidal_ideation | binary | Suicidal ideation |
| hiv_risk | binary | HIV risk from sexual violence |
| sought_help | binary | Sought help |
| help_source | categorical | health / police / family / community / ngo / religious |
| barrier_to_help | categorical | stigma / fear / normalisation / financial / no_awareness / distance |
| pep_offered | binary | PEP offered |
| pep_completed | binary | PEP completed |
| counselling_received | binary | Counselling |
| forensic_exam_done | binary | Forensic exam |
| police_report_filed | binary | Police report |

## 4. Validation

<p align="center">
<img src="validation_report.png" alt="Validation Report" width="100%">
</p>

Key validation checks:

- **IPV prevalence**: ~60% (any type), physical ~40% ✓
- **Help-seeking gradient**: 26% → 20% → 10% ✓
- **Depression in IPV+**: ~40% vs ~0% in non-IPV ✓
- **Partner alcohol**: Higher among IPV+ ✓
- **Barriers**: Stigma and fear dominant ✓
- **Clinical cascade drops**: PEP completion very low ✓

## 5. Usage

```python
from datasets import load_dataset
dataset = load_dataset("electricsheepafrica/gender-based-violence-ipv", "district_health_facility")
df = dataset["train"].to_pandas()
```

## 6. Limitations

- **Simulated**: Not from real GBV registries or DHS.
- **Women only**: Male victimisation not modelled.
- **No perpetrator data**: No detailed perpetrator characteristics.
- **Snapshot**: No longitudinal tracking of violence trajectories.
- **Sensitive topic**: Must be used with ethical consideration.

## 7. References

1. WHO (2024). Violence against women prevalence estimates.
2. Frontiers Public Health (2023). IPV disparities across 26 SSA countries.
3. PMC (2020). GBV in SSA systematic review.
4. UNFPA ESARO. Gender-based violence in East/Southern Africa.
5. PMC (2024). DHS 2019-2024 domestic violence determinants.
6. PubMed (2024). IPV in pregnancy South Africa.
7. PubMed (2010). Barriers to PEP completion after rape.

## Citation

```bibtex
@dataset{esa_gbv_ipv_2025,
title={Gender-Based Violence and IPV Dataset},
author={Electric Sheep Africa},
year={2025},
publisher={Hugging Face},
url={https://huggingface.co/datasets/electricsheepafrica/gender-based-violence-ipv}
}
```

## License

[CC-BY-4.0](https://creativecommons.org/licenses/by/4.0/)