Chunk #141 — Features and Pitfalls — Randomness and Stability

Source

An introduction to recursive partitioning: rationale, application, and characteristics of classification and regression trees, bagging, and random forests.

Embedded

yes

Text

Note that the two tuning parameters, ntree and mtry, also interact: To assess a high number of predictor variables in a data set, a high number of trees or a high number of preselected variables for each split, or ideally both, are necessary so that each variable has a chance to occur in enough trees. Only then its average variable importance measure is based on enough trials to actually reflect the importance of the variable and not just a random fluctuation.