Inverse Probability Weighting
Overview
Inverse probability weighting uses estimated probabilities of treatment, exposure, or inclusion to reweight observed data toward a target population or policy. It is used for causal effect estimation, off-policy evaluation in recommendation and bandits, and correction for missing or selectively labeled data. Common approaches include normalized or stabilized weights, propensity-score calibration, and augmented or doubly robust estimators that combine weighting with outcome predictions. A central challenge is that rare events create extreme weights, while poor overlap, misspecified propensities, and unobserved confounding can undermine accuracy and valid uncertainty estimates.
Latest papers
21 papers2 papers in the last four weeks, against 1 the four weeks before. 0.0% of all new papers.