Skip to main content
Browse Subject Areas

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here.

< Back to Article

Disaggregating Census Data for Population Mapping Using Random Forests with Remotely-Sensed and Ancillary Data

Fig 6

For Kenya we compared the summed predictions of population maps estimated using coarse census data (Level 4, “Division” level) to population counts at Level 5 (finer scale) units.

We present prediction errors (observed minus predicted) as percentage values of the observed Level 5 census counts. The RF approach results in far fewer census units with extreme over predictions (negative percent residuals in yellows and oranges) and under predictions (positive percent residuals in dark blues).

Fig 6