README.md

---
dataset:
- name: TemporalNeighborhoodMaterialWealthAfrica
tags:
  - satellite-imagery
  - wealth-index
  - development-economics
  - africa
  - socio-economic-data
  - deep-learning
configs:
- config_name: default
  data_files:
  - split: train
    path: wealth_africa.parquet # Default 'train' split in the viewer
license: bigscience-openrail-m
---

# Temporal Neighborhood-Level Material Wealth Maps of Africa (1990–2019)

This repository provides neighborhood-level material wealth estimates across Africa for the period 1990–2019. The data are stored in a single GeoTIFF file (`wealth_map.tif`), where each band corresponds to a three-year interval. These estimates were generated using a deep-learning model trained on Demographic and Health Surveys (DHS) data, as described in [Pettersson et al. (2023)](https://www.ijcai.org/proceedings/2023/0684.pdf).

## Overview

- **Data File**: `wealth_map.tif`  
- **Spatial Resolution**: ~6.72 km x 6.72 km  
- **Geographic Coverage**: Africa  
- **Temporal Coverage**: 1990–2019 (in 3-year intervals)  
- **Measurement Unit**: International Wealth Index (IWI), scaled from 0 to 1  
- **File Size**: ~52.2 MB  
- **MD5 Checksum**: `ab33e78dceeae49c06e753f0bb7eb904`  

### Bands and Time Periods

| **Band** | **Time Window** |
|:-------:|:---------------:|
| 1  | 1990–1992 |
| 2  | 1993–1995 |
| 3  | 1996–1998 |
| 4  | 1999–2001 |
| 5  | 2002–2004 |
| 6  | 2005–2007 |
| 7  | 2008–2010 |
| 8  | 2011–2013 |
| 9  | 2014–2016 |
| 10 | 2017–2019 |

## Description

These maps estimate the International Wealth Index (IWI) at a neighborhood resolution of approximately 6.72 km for all populated areas in Africa, as determined by the Global Human Settlement Layer (GHSL). The IWI is scaled between 0 and 1, representing a relative wealth measure derived from satellite imagery (Landsat, DMSP, and VIIRS). 

For further methodological details, please see:

> **Pettersson, M. B., Kakooei, M., Ortheden, J., Johansson, F. D., & Daoud, A. (2023).**  
> *Time Series of Satellite Imagery Improve Deep Learning Estimates of Neighborhood-Level Poverty in Africa.*  
> Proceedings of the Thirty-Second International Joint Conference on Artificial Intelligence (IJCAI 2023), 6165–6173.  
> [doi:10.24963/ijcai.2023/684](https://doi.org/10.24963/ijcai.2023/684)

Note that these data are also used in: [arxiv.org/abs/2508.01341](https://arxiv.org/abs/2508.01341). For more information, see the website for the AI & Global Development Lab: [**aidevlab.org**](https://aidevlab.org/).

## Additional Metadata

- **Deposit Date**: 2023-10-20  
- **Metadata Release Date**: 2023-10-20  
- **Publication Date**: 2023-10-20  
- **Type**: TIFF Image  
- **Description**: Multiband GeoTIFF containing IWI estimates for 10 time windows between 1990 and 2019.  

## How to Use

### Quick Start in Python

```python
import rasterio
import numpy as np

# Open the dataset
with rasterio.open("wealth_map.tif") as src:
    # Read band 1 (1990–1992)
    band1 = src.read(1)
    # Read band 10 (2017–2019)
    band10 = src.read(10)
    
    # Print basic info
    print("Band 1 shape:", band1.shape)
    print("Band 10 shape:", band10.shape)
    
    # Example: compute the mean wealth in 2017–2019
    mean_wealth_2017_2019 = np.nanmean(band10)
    print("Mean IWI (2017–2019):", mean_wealth_2017_2019)
```

## Additional Tabular Data: `poverty_improvement_by_state.csv`

This CSV file provides an aggregate measure of how average wealth has changed between the early 1990s and the late 2010s at the first-level administrative region (state/province) across Africa. Each row corresponds to a specific country–province pair, along with the estimated improvement in wealth over this period.

### Columns 

`Country`: Name of the country. 
`Province`: Name of the first-level administrative region (e.g., state or province). 
`Rank`: Ordering from largest to smallest improvement (1 indicates the greatest improvement). 
`Improvement`: Estimated change in the mean International Wealth Index (IWI) between 1990–1992 and 2017–2019 for that province.

### Data Source

Derived from the same deep-learning model as the main dataset (`wealth_map.tif`), as described in Pettersson et al. (2023). The IWI values for each province were averaged over the initial time window (1990–1992) and final time window (2017–2019). The difference of these two averages forms the Improvement value.

### Example Rows 

```
Country, Province, Rank, Improvement
Gambia, Banjul, 1, 0.31955
Mali, Bamako, 2, 0.31376
Botswana, Francistown, 3, 0.28345
Egypt, Kafr ash Shaykh, 4, 0.27021
Algeria, Boumerdès, 5, 0.26634
Egypt, Ash Sharqiyah, 6, 0.26594
Egypt, Bur Said, 7, 0.26321
Egypt, Ad Daqahliyah, 8, 0.26129
```

This table can help users quickly identify which provinces experienced the most significant gains in material wealth (as measured by IWI) over the nearly three-decade span. It complements the raster dataset by offering a province-level summary of changes in living conditions.

# Reference

**Pettersson, M. B., Kakooei, M., Ortheden, J., Johansson, F. D., & Daoud, A. (2023).** *Time Series of Satellite Imagery Improve Deep Learning Estimates of Neighborhood-Level Poverty in Africa.* Proceedings of the Thirty-Second International Joint Conference on Artificial Intelligence (IJCAI 2023), 6165–6173. [doi:10.24963/ijcai.2023/684](https://doi.org/10.24963/ijcai.2023/684)


```
@inproceedings{pettersson2023time,
  title={Time Series of Satellite Imagery Improve Deep Learning Estimates of Neighborhood-Level Poverty in Africa.},
  author={Pettersson, Markus B and Kakooei, Mohammad and Ortheden, Julia and Johansson, Fredrik D and Daoud, Adel},
  booktitle={IJCAI},
  pages={6165--6173},
  year={2023}
}
```


# Disclaimer

While we have made every effort to ensure the accuracy and reliability of these wealth estimates, they should be interpreted within the context and limitations of the source data and modeling methods. The authors and contributors accept no liability for any loss or damage arising from the use of this data.