electricsheepafrica/africa-conflict-somalia

---
annotations_creators:
- no-annotation
language_creators:
- found
language:
- en
license: cc-by-4.0
multilinguality:
- monolingual
size_categories:
- n<1K
source_datasets:
- original
task_categories:
- tabular-classification
- other
task_ids: []
tags:
- africa
- humanitarian
- hdx
- electric-sheep-africa
- som
pretty_name: "Baseline Assessment on Land Ownership, Land Rights, and Land Conflict in Somaliland"
dataset_info:
splits:
- name: train
num_examples: 410
- name: test
num_examples: 102
---

# Baseline Assessment on Land Ownership, Land Rights, and Land Conflict in Somaliland

**Publisher:** Observatory of Conflict and Violence Prevention (inactive) · **Source:** [HDX](https://data.humdata.org/dataset/baseline-assessment-on-land-ownership-land-rights-and-land-conflict-in-somaliland) · **License:** `cc-by-igo` · **Updated:** 2023-03-03

---

## Abstract

Due to the growing prevalence of land disputes in Somaliland, comprehensive knowledge of the local perspectives of land ownership, rights, and conflict is needed for effective programming and policy development. Observatory of Conflict and Violence Prevention has conducted a baseline assessment in order to gain a better understanding of the factors and dynamics of issues surrounding land ownership, land rights and land conflict in Somaliland.
Data was collected in December 2013 among 513 residents in Hargeisa, Gabiley, Borama, Salaxley, and Oodweyne districts. Focus group discussions and key informant interviews were also conducted within each district. Funding for this project was provided by the Japan Centre for Conflict Prevention (JCCP).

Each row in this dataset represents time-series observations. Temporal coverage is indicated by the `id_date`, `id_start` column(s). Geographic scope: **SOM**.

*Curated into ML-ready Parquet format by [Electric Sheep Africa](https://huggingface.co/electricsheepafrica).*

---

## Dataset Characteristics

| | |
|---|---|
| **Domain** | Conflict and security |
| **Unit of observation** | Time-series observations |
| **Rows (total)** | 513 |
| **Columns** | 96 (83 numeric, 8 categorical, 5 datetime) |
| **Train split** | 410 rows |
| **Test split** | 102 rows |
| **Geographic scope** | SOM |
| **Publisher** | Observatory of Conflict and Violence Prevention (inactive) |
| **HDX last updated** | 2023-03-03 |

---

## Variables

**Temporal** — `id_date`, `id_enddate`, `id_time` (range 4.0–73.0).

**Demographic** — `id_language` (range 1.0–2.0).

**Identifier / Metadata** — `id_format` (range 3.0–3.0), `id_start`, `id_end`, `id_interviewer` (ocvp3, ocvp2, ocvp1), `esa_source` (HDX) and 1 others.

**Other** — `serial` (range 1.0–671.0), `a1` (Mustafe, Asmahan, Hussein), `a2` (range 41.0–4059999.0), `a3`, `b1` (Hargeisa, Daad madheedh, Salaxley) and 81 others.

---

## Quick Start

```python
from datasets import load_dataset

ds = load_dataset("electricsheepafrica/africa-conflict-somalia")
train = ds["train"].to_pandas()
test = ds["test"].to_pandas()

print(train.shape)
train.head()
```

---

## Schema

| Column | Type | Null % | Range / Sample Values |
|---|---|---|---|
| `serial` | int64 | 0.0% | 1.0 – 671.0 (mean 286.9376) |
| `id_format` | int64 | 0.0% | 3.0 – 3.0 (mean 3.0) |
| `id_language` | int64 | 0.0% | 1.0 – 2.0 (mean 1.7505) |
| `id_date` | datetime64[ns] | 0.0% | |
| `id_start` | datetime64[ns] | 0.0% | |
| `id_enddate` | datetime64[ns] | 0.0% | |
| `id_end` | datetime64[ns] | 0.0% | |
| `id_time` | int64 | 0.0% | 4.0 – 73.0 (mean 12.5302) |
| `id_interviewer` | object | 0.0% | ocvp3, ocvp2, ocvp1 |
| `a1` | object | 0.0% | Mustafe, Asmahan, Hussein |
| `a2` | int64 | 0.0% | 41.0 – 4059999.0 (mean 135198.4113) |
| `a3` | datetime64[ns] | 0.0% | |
| `b1` | object | 0.0% | Hargeisa, Daad madheedh, Salaxley |
| `b2` | int64 | 0.0% | 1.0 – 5.0 (mean 2.9766) |
| `b3` | object | 4.5% | Hodan, Ayaan, Barwaaqo |
| `b4` | float64 | 25.0% | 1.0 – 2.0 (mean 1.0052) |
| `rp1` | int64 | 0.0% | 1.0 – 2.0 (mean 1.5848) |
| `rp2` | int64 | 0.0% | 16.0 – 999.0 (mean 51.501) |
| `rp3` | int64 | 0.0% | 1.0 – 4.0 (mean 1.8986) |
| `rp4` | int64 | 0.0% | 1.0 – 8.0 (mean 3.2164) |
| `lo1` | int64 | 0.0% | 1.0 – 2.0 (mean 1.4094) |
| `lo2` | float64 | 40.9% | 1.0 – 9.0 (mean 2.802) |
| `lo3` | int64 | 0.0% | 1.0 – 9.0 (mean 2.7076) |
| `lo4` | int64 | 0.0% | 1.0 – 7.0 (mean 3.0117) |
| `lo5` | int64 | 0.0% | 1.0 – 6.0 (mean 2.2066) |
| `lo6` | int64 | 0.0% | 1.0 – 7.0 (mean 1.8869) |
| `lo7` | int64 | 0.0% | 1.0 – 3.0 (mean 1.5712) |
| `lo8` | float64 | 54.8% | 1.0 – 5.0 (mean 2.0991) |
| `lo9` | int64 | 0.0% | 1.0 – 4.0 (mean 1.2164) |
| `lo10` | float64 | 20.5% | |
| `lo11` | int64 | 0.0% | |
| `lo12` | float64 | 79.3% | |
| `lo13` | int64 | 0.0% | |
| `lo14` | float64 | 3.3% | |
| `lra1` | int64 | 0.0% | |
| `lra2_1` | float64 | 6.8% | |
| `lra2_2` | float64 | 6.8% | |
| `lra2_3` | float64 | 6.8% | |
| `lra2_4` | float64 | 6.8% | |
| `lra2_5` | float64 | 6.8% | |
| `lra2_6` | float64 | 6.8% | |
| `lra2_7` | float64 | 6.8% | |
| `lra2_8` | float64 | 6.8% | |
| `lra2_9` | float64 | 6.8% | |
| `lra2_10` | float64 | 6.8% | |
| `lra2_11` | float64 | 6.8% | |
| `lra2_12` | float64 | 6.8% | |
| `lra4_1` | int64 | 0.0% | |
| `lra4_2` | int64 | 0.0% | |
| `lra4_3` | int64 | 0.0% | |
| `lra4_4` | int64 | 0.0% | |
| `lra4_5` | int64 | 0.0% | |
| `lra4_6` | int64 | 0.0% | |
| `lra4_7` | int64 | 0.0% | |
| `lra4_8` | int64 | 0.0% | |
| `lra5` | int64 | 0.0% | |
| `lra6` | float64 | 64.5% | |
| `lra9` | int64 | 0.0% | |
| `lc1` | int64 | 0.0% | |
| `lc2` | float64 | 21.1% | |
| `lc3` | float64 | 79.3% | |
| `lc4` | float64 | 79.3% | |
| `lc5` | object | 79.3% | sallaxley, ood wayne, barwaaqo |
| `lc6` | float64 | 79.3% | |
| `lc7` | float64 | 79.3% | |
| `lc8` | float64 | 79.3% | |
| `lc9` | int64 | 0.0% | |
| `lc10` | int64 | 0.0% | |
| `lc11` | int64 | 0.0% | |
| `lc12` | int64 | 0.0% | |
| `lc13` | object | 0.0% | 777, 999, bali ciise |
| `lc15_1` | int64 | 0.0% | |
| `lc15_2` | int64 | 0.0% | |
| `lc15_3` | int64 | 0.0% | |
| `lc15_4` | int64 | 0.0% | |
| `lc15_5` | int64 | 0.0% | |
| `lc15_6` | int64 | 0.0% | |
| `lc15_7` | int64 | 0.0% | |
| `lc15_8` | int64 | 0.0% | |
| `lc15_9` | int64 | 0.0% | |
| `lc16_1` | int64 | 0.0% | |
| `lc16_2` | int64 | 0.0% | |
| `lc16_3` | int64 | 0.0% | |
| `lc16_4` | int64 | 0.0% | |
| `lc16_5` | int64 | 0.0% | |
| `lc16_6` | int64 | 0.0% | |
| `lc16_7` | int64 | 0.0% | |
| `lc17_1` | int64 | 0.0% | |
| `lc17_2` | int64 | 0.0% | |
| `lc17_3` | int64 | 0.0% | |
| `lc17_4` | int64 | 0.0% | |
| `lc17_5` | int64 | 0.0% | |
| `lc17_6` | int64 | 0.0% | |
| `lc17_7` | int64 | 0.0% | |
| `esa_source` | object | 0.0% | HDX |
| `esa_processed` | object | 0.0% | 2026-04-27 |

---

## Numeric Summary

| Column | Min | Max | Mean | Median |
|---|---|---|---|---|
| `serial` | 1.0 | 671.0 | 286.9376 | 287.0 |
| `id_format` | 3.0 | 3.0 | 3.0 | 3.0 |
| `id_language` | 1.0 | 2.0 | 1.7505 | 2.0 |
| `id_time` | 4.0 | 73.0 | 12.5302 | 11.0 |
| `a2` | 41.0 | 4059999.0 | 135198.4113 | 762.0 |
| `b2` | 1.0 | 5.0 | 2.9766 | 3.0 |
| `b4` | 1.0 | 2.0 | 1.0052 | 1.0 |
| `rp1` | 1.0 | 2.0 | 1.5848 | 2.0 |
| `rp2` | 16.0 | 999.0 | 51.501 | 35.0 |
| `rp3` | 1.0 | 4.0 | 1.8986 | 2.0 |
| `rp4` | 1.0 | 8.0 | 3.2164 | 3.0 |
| `lo1` | 1.0 | 2.0 | 1.4094 | 1.0 |
| `lo2` | 1.0 | 9.0 | 2.802 | 2.0 |
| `lo3` | 1.0 | 9.0 | 2.7076 | 2.0 |
| `lo4` | 1.0 | 7.0 | 3.0117 | 3.0 |

---

## Curation

Raw data was downloaded from HDX via the CKAN API and converted to Parquet. Column names were lowercased and standardised to snake_case. Common missing-value markers (`N/A`, `null`, `none`, `-`, `unknown`, `no data`, `#N/A`) were unified to `NaN`. 30 column(s) with >80% missing values were removed: `l02o`, `lo3o`, `lo13o`, `lo14o`, `lra2o`, `lra3_1`.... 5 column(s) were cast from string to numeric or datetime based on parse-success rate (>85% threshold). The dataset was split 80/20 into train and test partitions using a fixed random seed (42) and saved as Snappy-compressed Parquet.

---

## Limitations

- Data originates from Observatory of Conflict and Violence Prevention (inactive) and has not been independently validated by ESA.
- Automated cleaning cannot correct for misreported values, definitional inconsistencies, or sampling bias in the original collection.
- The following columns have >20% missing values and should be treated with caution in modelling: `b4`, `lo2`, `lo8`, `lo10`, `lo12`, `lra6`, `lc2`, `lc3`....
- Refer to the [original HDX dataset page](https://data.humdata.org/dataset/baseline-assessment-on-land-ownership-land-rights-and-land-conflict-in-somaliland) for the publisher's own methodology notes and caveats.

---

## Citation

```bibtex
@dataset{hdx_africa_conflict_somalia,
title = {Baseline Assessment on Land Ownership, Land Rights, and Land Conflict in Somaliland},
author = {Observatory of Conflict and Violence Prevention (inactive)},
year = {2023},
url = {https://data.humdata.org/dataset/baseline-assessment-on-land-ownership-land-rights-and-land-conflict-in-somaliland},
note = {Repackaged for machine learning by Electric Sheep Africa (https://huggingface.co/electricsheepafrica)}
}
```

---

*[Electric Sheep Africa](https://huggingface.co/electricsheepafrica) — Africa's ML dataset infrastructure. Lagos, Nigeria.*