We request your patience during the alpha release of the Global Poverty Map, which currently only supports Google Chrome, Safari, and Firefox browsers. The full release will have complete browser compatibility. Email didl@berkeley.edu with questions or concerns.

This map shows estimates of wealth and poverty around the world. The estimates are computed from satellite imagery, internet data, and other non-traditional sources of data.

The height and color of each grid cell help visualize micro-regional wealth. By default, cell height indicates absolute wealth (relative to the world), and cell color indicates relative wealth (relative to other regions of the country).

You can choose different metrics to display by clicking on the settings icon. The default view simultaneously displays the Absolute Wealth Estimate (using cell height) and the Relative Wealth Index (using cell color) of each location. This view (AWE+confidence) instead uses cell color to indicate the confidence of the estimate at each grid cell.

Explore the map yourself by clicking and dragging, by using the scroll wheel to zoom, and by using the right mouse button (or holding CTRL) to tilt or rotate the point of view.

AWE + RWI
AWE: MAX
RWI: >1.0
AWE: 1/2 MAX
RWI: 0.5
AWE: 0
RWI: <-1.0
AWE: 1/2 MAX
RWI: 0

Absolute Wealth

Height

higher

lower

Relative
Index

Color

AWE + Confidence
AWE: MAX
RWI: >1.0
AWE: 1/2 MAX
RWI: 0.5
AWE: 0
RWI: <-1.0
AWE: 1/2 MAX
RWI: 0

Absolute Wealth

Height

higher

lower

Confidence
Index

Color

Key: Relative Wealth Display
AWE: MAX
RWI: >1.0
AWE: 1/2 MAX
RWI: 0.5
AWE: 0
RWI: <-1.0
AWE: 1/2 MAX
RWI: 0

higher

lower

Relative
Index

Color

Confidence

lesser

greater

Opacity

Mean Confidence Index: 0.75
Display Mode
AWE+RWI
AWE+Confidence
RWI+Confidence

AWE+RWI simultaneously displays the Absolute Wealth Estimate (using cell height) and the Relative Wealth Index (using cell color) of each location. Refer to our paper to learn more about what these numbers mean.

AWE+Confidence simultaneously displays the Absolute Wealth Estimate (using cell height) and the confidence of the estimate (using cell color) of each location. Refer to our paper to learn more about what these numbers mean.

RWI+Confidence simultaneously displays the Relative Wealth Index (using cell color) and the confidence of the estimate (using transparency) of each location. Refer to our paper to learn more about what these numbers mean.

Data Scale

By default, the height of the bars auto-adjusts to the region being displayed. By setting the height manually, you can compare the height of two different locations using the same scale.

Auto Manual
1
100
AWE + RWI
AWE: MAX
RWI: >1.0
AWE: 1/2 MAX
RWI: 0.5
AWE: 0
RWI: <-1.0
AWE: 1/2 MAX
RWI: 0

Absolute Wealth

Height

higher

lower

Relative
Index

Color

AWE + Confidence
AWE: MAX
RWI: >1.0
AWE: 1/2 MAX
RWI: 0.5
AWE: 0
RWI: <-1.0
AWE: 1/2 MAX
RWI: 0

Absolute Wealth

Height

higher

lower

Confidence
Index

Color

RWI + Confidence
AWE: MAX
RWI: >1.0
AWE: 1/2 MAX
RWI: 0.5
AWE: 0
RWI: <-1.0
AWE: 1/2 MAX
RWI: 0

higher

lower

Relative
Index

Color

Confidence

lesser

greater

Opacity

Data Confidence

The mean confidence index indicates the average confidence of the wealth estimates, averaged over the entire region shown in the current view.

Mean Confidence Index: 0.75

The data sources include satellites, mobile phone networks, social media, and other geospatial data. These “big” data sources are matched to the “traditional” survey data using geographic markers present in both datasets.

Nigeria

Villages with surveys
(N=899)

Geospatial Big Data

Satellites hi-res imagery, night lights
Connectivity cell towers, devices
Demography population, urban/rural
Geography road density, elevation

Ground Truth Data

20-50 Surveys Per Village
2-4 Hours Per Survey

The data sources include satellites, mobile phone networks, social media, and other geospatial data. These “big” data sources are matched to the “traditional” survey data using geographic markers present in both datasets.

Nigeria

Villages with surveys
(N=899)

Geospatial Big Data

Satellites hi-res imagery, night lights
Connectivity cell towers, devices
Demography population, urban/rural
Geography road density, elevation

Ground Truth Data

20-50 Surveys Per Village
2-4 Hours Per Survey

These algorithms “learn” which characteristics of the big data are useful in predicting wealth. For instance, villages with good cell phone coverage, and with lots of paved surface, tend to be wealthier.

Geospatial Big Data

Satellites
Connectivity
Demography
Geography

Ground Truth Data

20-50 Surveys Per Village
2-4 Hours Per Survey

Supervised Machine Learning

Trained Model

“Signatures” of poverty in geospatial big data
Model to predict poverty in regions without ground truth

Once the machine learning model is trained and validated using ground truth data (i.e., once we ensure that the model can accurately predict the wealth of villages where we know the wealth values), the model is used to predict the wealth of regions where there is no survey data.

Absolute Wealth

MAX

0

Height

higher

lower

Relative
Index

MAX

MIN

Color

Confidence
Index

1

0

Color

Mean Confidence:

= 20 km