Reproduce “Puffin: Probabilistic Learning of Spatial Detail From Coarse Observations” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Chaitanya Jobanputra, Sebastian Vollmer, Gerrit Großmann
PUFFIN disaggregates coarse counts into smaller areas using satellite embeddings while preserving the observed regional total. Its probabilistic model predicts subregion counts from regional totals alone, but calibrated uncertainty intervals require fine-resolution validation labels. In Germany’s census-grid experiment, its within-region R-squared averaged 0.561 across 18 configurations, versus 0.544 for Goodman-style regression; PUFFIN won 16 configurations, with margins no larger than 0.04.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.