Track papers across labs, topics, and trends.
Search papers, labs, and topics across Lattice.
Puffin is a probabilistic framework that statistically disaggregates coarse socioeconomic data to higher resolutions by leveraging high-resolution satellite embeddings as covariates. It learns a probability distribution for each subregion via an aggregation-aware likelihood, enabling conditional predictions that are consistent with observed regional totals and provide calibrated uncertainty estimates without requiring fine-resolution ground truth for training. Evaluations on German and US census data demonstrate Puffin's effectiveness across various count variables and provide insights into disaggregation success factors.
Satellite imagery unlocks fine-grained socioeconomic predictions with calibrated uncertainty, even without direct ground truth.
High-resolution socioeconomic variables are important for applications such as urban planning, public health, disaster response, and resource allocation. In practice, however, these variables are often observed only at a coarse spatial resolution. We introduce Puffin, a probabilistic framework for statistical disaggregation that raises the resolution of coarse totals using high-resolution satellite embeddings as covariates. Instead of predicting a single value for each fine-resolution subregion, Puffin learns a probability distribution and is trained through an aggregation-aware likelihood. At inference, Puffin conditions these predictions on the observed regional total and splits it among the subregions. The resulting fine-scale estimates are consistent with the observed aggregate and come with calibrated uncertainty, without requiring fine-resolution labels for training. We evaluate Puffin on German and US census, employment, and election data across population, jobs, and other count variables, and study when statistical disaggregation succeeds or fails across regions, countries, and targets.