Propensity score matching
Statistical matching technique
📊 Rating
1 news mentions · 👍 0 likes · 👎 0 dislikes
📌 Topics
- Artificial Intelligence (1)
- Data Science (1)
- Machine Learning (1)
🏷️ Keywords
Representation learning (1) · Data heterogeneity (1) · Zero-shot generalization (1) · Propensity score matching (1) · Algorithm bias (1) · arXiv (1) · Machine learning models (1)
📖 Key Information
In the statistical analysis of observational data, propensity score matching (PSM) is a statistical matching technique that attempts to estimate the effect of a treatment, policy, or other intervention by accounting for the covariates that predict receiving the treatment. PSM attempts to reduce the bias due to confounding variables that could be found in an estimate of the treatment effect obtained from simply comparing outcomes among units that received the treatment versus those that did not.
Paul R. Rosenbaum and Donald Rubin introduced the technique in 1983, defining the propensity score as the conditional probability of a unit (e.g., person, classroom, school) being assigned to the treatment, given a set of observed covariates.
📰 Related News (1)
-
🇺🇸 Beyond Pooling: Matching for Robust Generalization under Data Heterogeneity
arXiv:2602.07154v1 Announce Type: cross Abstract: Pooling heterogeneous datasets across domains is a common strategy in representation learning, but ...
🔗 Entity Intersection Graph
People and organizations frequently mentioned alongside Propensity score matching:
- 🌐 Feature learning (1 shared articles)