Leo Breiman
University of California, Berkeley
Department of Statistics, Emeritus
Founder, California Statistical Software, Inc.
Leo Breiman, one of the four creators of CART®, was elected to the National Academy of Sciences on May 1, 2001 for his seminal work in data mining, decision trees, and pattern recognition. Election to the Academy is considered one of the highest honors that can be accorded to a scientist or engineer. Leo joins CART co-author Charles J. Stone who was elected earlier.
Here is the full text of the NAS citation:
Breiman has done fundamental work in stochastic processes, information theory, and mathematical statistics. He is a seminal thinker who has developed modern methods of classification and pattern recognition. He has made significant contributions to the practice of statistics bridging the gaps between that field, signal processing, and computer science. The CART monograph, Classification and Regression Trees, authored by Leo Breiman, Jerome H. Friedman, Richard Olshen, and Charles J. Stone is available from CRC Press. Click here for further information on this book.
We have converted several of Leo's papers on data mining with decision trees to PDF format for download from this site:
- Bagging Predictors
The classic article introducing bootstrap resampling as a method for generating an ensemble of multiple trees that vote to predict or classify. - ARCing Classifiers
Another classic paper introducing Adaptive Resampling and Combining (a variant of boosting). - Bias, Variance, and ARCing Classifiers
An exploration into why bagging and boosting improve prediction accuracy. - Some Infinity Theory for Predictor Ensembles
Technical exploration into why ensembles work.