electricsheepafrica/africa-global-mpi

Global Multidimensional Poverty Index

Publisher: Oxford Poverty & Human Development Initiative · Source: HDX · License: other-pd-nr · Updated: 2026-03-05

---

Abstract

The global Multidimensional Poverty Index provides the only comprehensive measure available for non-income poverty, which has become a critical underpinning of the SDGs. The global Multidimensional Poverty Index (MPI) measures multidimensional poverty in over 100 developing countries, using internationally comparable datasets and is updated annually. The measure captures the acute deprivations that each person faces at the same time using information from 10 indicators, which are grouped into three equally weighted dimensions: health, education, and living standards. Critically, the MPI comprises variables that are already reported under the Demographic Health Surveys (DHS), the Multi-Indicator Cluster Surveys (MICS) and in some cases, national surveys.

The subnational multidimensional poverty data from the data tables are published by the Oxford Poverty and Human Development Initiative (OPHI), University of Oxford. For the details of the global MPI methodology, please see the latest Methodological Notes found here.

Each row in this dataset represents country-level aggregates. Data was last updated on HDX on 2026-03-05. Geographic scope: AFG, ALB, DZA, AGO, ARG, ARM, BGD, BRB, and 104 others.

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

---

Dataset Characteristics

Domain	Public health
Unit of observation	Country-level aggregates
Rows (total)	1,468
Columns	13 (5 numeric, 6 categorical, 0 datetime)
Train split	1,174 rows
Test split	293 rows
Geographic scope	AFG, ALB, DZA, AGO, ARG, ARM, BGD, BRB, and 104 others
Publisher	Oxford Poverty & Human Development Initiative
HDX last updated	2026-03-05

---

Variables

Geographic — country_iso3 (KEN, NGA, IND), admin_1_pcode (HT04, TN3, GH08), admin_1_name (Central, Western, Eastern), intensity_of_deprivation (range 33.3333–69.6577), vulnerable_to_poverty (range 0.0–48.3534) and 2 others.

Temporal — start_date, end_date.

Outcome / Measurement — headcount_ratio (range 0.0–99.2537).

Identifier / Metadata — esa_source (HDX), esa_processed (2026-04-04).

Other — mpi (range 0.0–0.6759).

---

Quick Start

from datasets import load_dataset

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

print(train.shape)
train.head()

---

Schema

Column	Type	Null %	Range / Sample Values
country_iso3	object	0.0%	KEN, NGA, IND
admin_1_pcode	object	40.7%	HT04, TN3, GH08
admin_1_name	object	7.4%	Central, Western, Eastern
mpi	float64	0.0%	0.0 – 0.6759 (mean 0.1482)
headcount_ratio	float64	0.0%	0.0 – 99.2537 (mean 29.2917)
intensity_of_deprivation	float64	1.2%	33.3333 – 69.6577 (mean 44.2109)
vulnerable_to_poverty	float64	0.0%	0.0 – 48.3534 (mean 15.0861)
in_severe_poverty	float64	0.0%	0.0 – 91.6007 (mean 14.5411)
survey	object	0.0%	DHS, MICS, PAPFAM
start_date	datetime64[ns, UTC]	0.0%
end_date	datetime64[ns, UTC]	0.0%
esa_source	object	0.0%	HDX
esa_processed	object	0.0%	2026-04-04

---

Numeric Summary

Column	Min	Max	Mean	Median
mpi	0.0	0.6759	0.1482	0.0743
headcount_ratio	0.0	99.2537	29.2917	17.418
intensity_of_deprivation	33.3333	69.6577	44.2109	42.5692
vulnerable_to_poverty	0.0	48.3534	15.0861	14.8096
in_severe_poverty	0.0	91.6007	14.5411	3.8831

---

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. 2 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 Oxford Poverty & Human Development Initiative 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: admin_1_pcode.
• This dataset spans 112 countries; geographic and methodological inconsistencies across national boundaries may affect cross-country comparability.
• Refer to the original HDX dataset page for the publisher's own methodology notes and caveats.

---

Citation

@dataset{hdx_africa_global_mpi,
  title     = {Global Multidimensional Poverty Index},
  author    = {Oxford Poverty & Human Development Initiative},
  year      = {2026},
  url       = {https://data.humdata.org/dataset/global-mpi},
  note      = {Repackaged for machine learning by Electric Sheep Africa (https://huggingface.co/electricsheepafrica)}
}

---

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