Fig. 4: Comparison of model performance on North Korea and least developed countries (LDCs) in Asia in terms of economic indicator prediction for the human-machine collaborative model (red, circle) and four other baselines: NL(nightlight)-regression (gray, cross-shaped), NL-guided POG (black, star-shaped), Land cover-guided POG (green, square), and Relative wealth index (RWI) (sky blue, triangle). | Nature Communications

Fig. 4: Comparison of model performance on North Korea and least developed countries (LDCs) in Asia in terms of economic indicator prediction for the human-machine collaborative model (red, circle) and four other baselines: NL(nightlight)-regression (gray, cross-shaped), NL-guided POG (black, star-shaped), Land cover-guided POG (green, square), and Relative wealth index (RWI) (sky blue, triangle).

From: A human-machine collaborative approach measures economic development using satellite imagery

Fig. 4

Based on the GDP per capita (at current prices USD, 2019), we select five countries that have a similar economic rank to North Korea (198th): Nepal (183rd), Myanmar (174th), Cambodia (169th), Bangladesh (164th), Laos (154th)28. The grid-level evaluation is based on the density of building area (North Korea) and population estimates (LDCs in Asia) as a proxy for economic development. The district-level performance is calculated using official statistics from census and surveys: density of building area (North Korea), population, establishments and employment (LDCs in Asia). We use a simple unweighted average for aggregating grid scores at the district level.

Back to article page