electricsheepafrica/africa-climate-change-impact-in-niger

---
annotations_creators:
- no-annotation
language_creators:
- found
language:
- en
license: cc-by-4.0
multilinguality:
- monolingual
size_categories:
- 1K<n<10K
source_datasets:
- original
task_categories:
- tabular-classification
- other
task_ids: []
tags:
- africa
- humanitarian
- hdx
- electric-sheep-africa
- affected-population
- climate-weather
- fatalities
- geodata
- humanitarian-needs-overview-hno
- ner
pretty_name: "Disaster Loss Data for Niger"
dataset_info:
splits:
- name: train
num_examples: 2960
- name: test
num_examples: 740
---

# Disaster Loss Data for Niger

**Publisher:** United Nations Office for Disaster Risk Reduction (UNDRR) · **Source:** [HDX](https://data.humdata.org/dataset/climate-change-impact-in-niger) · **License:** `cc-by-igo` · **Updated:** 2023-05-16

---

## Abstract

Number of Deaths, Injured, Missing, Houses Destroyed, Houses Damaged, Victims Affected, Relocated, Evacuated, Losses and Damages in crops by climate change event

Each row in this dataset represents first-level administrative unit observations. Temporal coverage is indicated by the `date_ymd` column(s). Geographic scope: **NER**.

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

---

## Dataset Characteristics

| | |
|---|---|
| **Domain** | Climate and environment |
| **Unit of observation** | First-level administrative unit observations |
| **Rows (total)** | 3,700 |
| **Columns** | 30 (21 numeric, 8 categorical, 1 datetime) |
| **Train split** | 2,960 rows |
| **Test split** | 740 rows |
| **Geographic scope** | NER |
| **Publisher** | United Nations Office for Disaster Risk Reduction (UNDRR) |
| **HDX last updated** | 2023-05-16 |

---

## Variables

**Geographic** — `code_region` (range 1.0–8.0), `region` (TILLABERI, TAHOUA, ZINDER), `location` (Tillabéry, Kollo, Filingué), `date_ymd`, `houses_destroyed` (range 0.0–3542.0).

**Demographic** — `houses_damaged` (range 0.0–218.0), `damages_in_crops_ha` (range 0.0–370000.0), `damages_in_roads_mts` (range 0.0–0.0).

**Outcome / Measurement** — `deaths` (range 0.0–666.0), `affected` (range 0.0–445361.0).

**Identifier / Metadata** — `code_department` (range 101.0–800.0), `code_commune` (range 10101.0–80005.0), `esa_source` (HDX), `esa_processed` (2026-04-17).

**Other** — `serial` (range 1.0–3700.0), `event` (EPIDEMIC, FLOOD, EPIZOOTY), `department` (TILLABERI, MADAROUNFA, KOLLO), `commune` (TILLABERI, MIRRIAH, DIFFA), `description_of_cause` (Rougeole, suite à une importante quantité de pluie tombée, manque de pluie) and 11 others.

---

## Quick Start

```python
from datasets import load_dataset

ds = load_dataset("electricsheepafrica/africa-climate-change-impact-in-niger")
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 – 3700.0 (mean 1848.7641) |
| `event` | object | 0.0% | EPIDEMIC, FLOOD, EPIZOOTY |
| `code_region` | int64 | 0.0% | 1.0 – 8.0 (mean 4.6811) |
| `region` | object | 0.0% | TILLABERI, TAHOUA, ZINDER |
| `code_department` | float64 | 2.9% | 101.0 – 800.0 (mean 474.3403) |
| `department` | object | 2.9% | TILLABERI, MADAROUNFA, KOLLO |
| `code_commune` | float64 | 8.4% | 10101.0 – 80005.0 (mean 47901.1209) |
| `commune` | object | 8.4% | TILLABERI, MIRRIAH, DIFFA |
| `location` | object | 37.9% | Tillabéry, Kollo, Filingué |
| `date_ymd` | datetime64[ns] | 0.5% | |
| `description_of_cause` | object | 52.9% | Rougeole, suite à une importante quantité de pluie tombée, manque de pluie |
| `datacards` | int64 | 0.0% | 1.0 – 1.0 (mean 1.0) |
| `deaths` | float64 | 0.1% | 0.0 – 666.0 (mean 2.8755) |
| `injured` | int64 | 0.0% | 0.0 – 139.0 (mean 0.1505) |
| `missing` | int64 | 0.0% | 0.0 – 0.0 (mean 0.0) |
| `houses_destroyed` | int64 | 0.0% | 0.0 – 3542.0 (mean 19.4422) |
| `houses_damaged` | int64 | 0.0% | 0.0 – 218.0 (mean 0.4389) |
| `victims` | int64 | 0.0% | 0.0 – 567954.0 (mean 4134.3024) |
| `affected` | int64 | 0.0% | 0.0 – 445361.0 (mean 373.7105) |
| `relocated` | int64 | 0.0% | 0.0 – 13146.0 (mean 16.1114) |
| `evacuated` | int64 | 0.0% | 0.0 – 0.0 (mean 0.0) |
| `losses_usd` | int64 | 0.0% | 0.0 – 59000.0 (mean 21.4865) |
| `losses_local` | int64 | 0.0% | 0.0 – 1794000000.0 (mean 623402.1108) |
| `education_centers` | int64 | 0.0% | 0.0 – 5.0 (mean 0.0054) |
| `hospitals` | int64 | 0.0% | 0.0 – 1.0 (mean 0.0003) |
| `damages_in_crops_ha` | float64 | 0.0% | 0.0 – 370000.0 (mean 711.8333) |
| `damages_in_roads_mts` | int64 | 0.0% | 0.0 – 0.0 (mean 0.0) |
| `lost_cattle` | int64 | 0.0% | |
| `esa_source` | object | 0.0% | HDX |
| `esa_processed` | object | 0.0% | 2026-04-17 |

---

## Numeric Summary

| Column | Min | Max | Mean | Median |
|---|---|---|---|---|
| `serial` | 1.0 | 3700.0 | 1848.7641 | 1849.5 |
| `code_region` | 1.0 | 8.0 | 4.6811 | 5.0 |
| `code_department` | 101.0 | 800.0 | 474.3403 | 505.0 |
| `code_commune` | 10101.0 | 80005.0 | 47901.1209 | 50605.0 |
| `datacards` | 1.0 | 1.0 | 1.0 | 1.0 |
| `deaths` | 0.0 | 666.0 | 2.8755 | 0.0 |
| `injured` | 0.0 | 139.0 | 0.1505 | 0.0 |
| `missing` | 0.0 | 0.0 | 0.0 | 0.0 |
| `houses_destroyed` | 0.0 | 3542.0 | 19.4422 | 0.0 |
| `houses_damaged` | 0.0 | 218.0 | 0.4389 | 0.0 |
| `victims` | 0.0 | 567954.0 | 4134.3024 | 0.0 |
| `affected` | 0.0 | 445361.0 | 373.7105 | 0.0 |
| `relocated` | 0.0 | 13146.0 | 16.1114 | 0.0 |
| `evacuated` | 0.0 | 0.0 | 0.0 | 0.0 |
| `losses_usd` | 0.0 | 59000.0 | 21.4865 | 0.0 |

---

## 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`. 5 column(s) with >80% missing values were removed: `cause`, `source`, `magnitude`, `glidenumber`, `other_sectors`. 1 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 United Nations Office for Disaster Risk Reduction (UNDRR) 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: `location`, `description_of_cause`.
- Refer to the [original HDX dataset page](https://data.humdata.org/dataset/climate-change-impact-in-niger) for the publisher's own methodology notes and caveats.

---

## Citation

```bibtex
@dataset{hdx_africa_climate_change_impact_in_niger,
title = {Disaster Loss Data for Niger},
author = {United Nations Office for Disaster Risk Reduction (UNDRR)},
year = {2023},
url = {https://data.humdata.org/dataset/climate-change-impact-in-niger},
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.*