electricsheepafrica/africa-somalia-operational-presence

Somalia: Operational Presence

Publisher: OCHA Somalia · Source: HDX · License: cc-by · Updated: 2026-04-03

---

Abstract

The Who does What Where (3W) is a core humanitarian coordination dataset. It is critical to know where humanitarian organizations are working, what they are doing and their capability in order to identify gaps, avoid duplication of efforts, and plan for future humanitarian response (if needed). The data includes a list of humanitarian organizations by district and cluster, as well as a unique count of organizations.

Each row in this dataset represents subnational administrative unit observations. Data was last updated on HDX on 2026-04-03. Geographic scope: SOM.

Curated into ML-ready Parquet format by Electric Sheep Africa.

---

Dataset Characteristics

Domain	Humanitarian and development data
Unit of observation	Subnational administrative unit observations
Rows (total)	9,684
Columns	24 (5 numeric, 19 categorical, 0 datetime)
Train split	7,747 rows
Test split	1,936 rows
Geographic scope	SOM
Publisher	OCHA Somalia
HDX last updated	2026-04-03

---

Variables

Geographic — activity_description (Raising awareness on the risks/threat posed explosive hazards, MRE sessions, Relief: General Food Distribution (GFD)), region (Bay, Mudug, Banadir), district (Baidoa, Gaalkacyo, Belet Weyne), specific_location (Baidoa, Kismayo, Belet Weyne), start_date_dd_mm_yy and 4 others.

Demographic — total_of_individuals (range 0.0–317122.7774), of_households.

Identifier / Metadata — programme_name (ICSP, TSFP, OTP), esa_source, esa_processed.

Other — cluster (Protection, Education, WASH), organization (WFP, RI, SCI), implementing_partner (RI, HALO Trust, SCI), donor (OFDA, OCHA, WFP), rural_urban (Rural, Urban, IDPs/Non IDPs) and 5 others.

---

Quick Start

from datasets import load_dataset

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

print(train.shape)
train.head()

---

Schema

Column	Type	Null %	Range / Sample Values
cluster	object	0.0%	Protection, Education, WASH
organization	object	0.0%	WFP, RI, SCI
implementing_partner	object	0.0%	RI, HALO Trust, SCI
programme_name	object	61.7%	ICSP, TSFP, OTP
activity_description	object	16.5%	Raising awareness on the risks/threat posed explosive hazards, MRE sessions, Relief: General Food Distribution (GFD)
donor	object	34.4%	OFDA, OCHA, WFP
region	object	0.0%	Bay, Mudug, Banadir
district	object	0.0%	Baidoa, Gaalkacyo, Belet Weyne
specific_location	object	2.0%	Baidoa, Kismayo, Belet Weyne
rural_urban	object	62.1%	Rural, Urban, IDPs/Non IDPs
status	object	0.1%
duration	object	65.3%
start_date_dd_mm_yy	object	1.9%
end_date_dd_mm_yy	object	1.4%
y_coord	object	63.2%
x_coord	object	63.1%
total_of_individuals	float64	20.4%	0.0 – 317122.7774 (mean 1316.2217)
of_households	object	65.3%
women	float64	33.1%	0.0 – 103579.63 (mean 486.0282)
men	float64	35.6%	0.0 – 107807.37 (mean 330.0472)
girls	float64	43.4%	0.0 – 50985.0 (mean 419.6685)
boys	float64	32.6%	0.0 – 54981.7587 (mean 10269.4661)
esa_source	object	0.0%
esa_processed	object	0.0%

---

Numeric Summary

Column	Min	Max	Mean	Median
total_of_individuals	0.0	317122.7774	1316.2217	64.0
women	0.0	103579.63	486.0282	31.0
men	0.0	107807.37	330.0472	17.0
girls	0.0	50985.0	419.6685	30.0
boys	0.0	54981.7587	10269.4661	80.0

---

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. 9 column(s) with >80% missing values were removed: modality, conditionality, restriction, transfer_value_usd, frequency, delivery_mechanism.... 6,055 exact duplicate rows were removed. 4 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 OCHA Somalia 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: programme_name, donor, rural_urban, duration, y_coord, x_coord, total_of_individuals, of_households....
• Refer to the original HDX dataset page for the publisher's own methodology notes and caveats.

---

Citation

@dataset{hdx_africa_somalia_operational_presence,
  title     = {Somalia: Operational Presence},
  author    = {OCHA Somalia},
  year      = {2026},
  url       = {https://data.humdata.org/dataset/somalia-operational-presence},
  note      = {Repackaged for machine learning by Electric Sheep Africa (https://huggingface.co/electricsheepafrica)}
}

---

Electric Sheep Africa — Africa's ML dataset infrastructure. Lagos, Nigeria.