When does Subagging Work?
This provides theoretical insights for practitioners using ensemble methods in machine learning, though it is incremental as it builds on known bias-variance trade-offs.
The paper analyzes subagging (subsample aggregating) on regression trees, showing that it improves performance over single trees for any number of splits, with greater gains for many splits, but a single optimally sized tree can outperform subagging if tree sizes are not chosen optimally.
We study the effectiveness of subagging, or subsample aggregating, on regression trees, a popular non-parametric method in machine learning. First, we give sufficient conditions for pointwise consistency of trees. We formalize that (i) the bias depends on the diameter of cells, hence trees with few splits tend to be biased, and (ii) the variance depends on the number of observations in cells, hence trees with many splits tend to have large variance. While these statements for bias and variance are known to hold globally in the covariate space, we show that, under some constraints, they are also true locally. Second, we compare the performance of subagging to that of trees across different numbers of splits. We find that (1) for any given number of splits, subagging improves upon a single tree, and (2) this improvement is larger for many splits than it is for few splits. However, (3) a single tree grown at optimal size can outperform subagging if the size of its individual trees is not optimally chosen. This last result goes against common practice of growing large randomized trees to eliminate bias and then averaging to reduce variance.