Overview
Leo Breiman (1928–2005) was a statistician (Professor, University of California, Berkeley) and one of the founders of modern statistical machine learning — creator of CART, bagging, and random forests, and author of the influential "Statistical Modeling: The Two Cultures" essay.
Key Contributions / Features
- Random forests (Breiman 2001): ensembles of decorrelated bootstrap-and-random-feature trees with a strength–correlation error bound and out-of-bag diagnostics (Random Forest).
- CART (Breiman-Friedman-Olshen-Stone 1984): the classification and regression tree framework (CART).
- Bagging (Breiman 1996): bootstrap aggregation for variance reduction — the precursor to random forests.
- "The Two Cultures" (2001): a manifesto for predictive/algorithmic modelling over data-generating-model statistics.
Related