Loading timeline…
20012000s
Leo Breiman
Berkeley statistician who created random forests and argued statistics should be judged on prediction, not on models.
Organizations
University of California, Berkeley
Major Achievements
- •Introduced random forests (2001), still one of the strongest general-purpose methods for tabular data.
- •Co-created CART (1984), the classification and regression tree framework behind decision-tree learning.
- •Developed bagging (1996), the bootstrap-aggregation idea underlying modern ensemble methods.
- •His 2001 essay 'Statistical Modeling: The Two Cultures' made the case for predictive accuracy over model interpretation.
Key Papers
- Random Forests
Leo Breiman · Machine Learning · 2001
- Statistical Modeling: The Two Cultures
Leo Breiman · Statistical Science · 2001