Dataset: A third of sub-Saharan Africa's farms cultivate less than half an hectare of land

Farm sizes are highly heterogeneous across sub-Saharan Africa, yet challenging to describe. Here, we standardize household survey data from 16 countries to predict through machine learning the empirical distribution of smallholder farm sizes in10 x 10 km2 grid cells across the continent. Key outputs include rasters (.tif) predicting the spatial distribution of average farm sizes and their associated centiles. A workflow is also proposed, which can be used/improved as more/better data are available.