Robust regression trees based on M-estimators

GALIMBERTI, GIULIANO; Department Of Statistics - University Of Bologna, PILLATI, MARILENA; Department Of Statistics - University Of Bologna & SOFFRITTI, GABRIELE; Department Of Statistics - University Of Bologna
The paper addresses the problem of robustness of regression trees with respect to outlying values in the dependent variable. New robust tree-based procedures are described, which are obtained by introducing in the tree building phase some objective functions already used in the linear robust regression approach, namely Huber’s and Tukey’s bisquare functions. The performance of the new procedures is evaluated through a Monte Carlo experiment.