Hierarchical Shrinkage: improving the accuracy and interpretability of tree-based methods

TL;DR

Hierarchical Shrinkage (HS) is a post-hoc regularization technique for tree-based models that improves accuracy and interpretability by shrinking node predictions towards ancestor means, without altering tree structure.

Contribution

The paper introduces Hierarchical Shrinkage, a fast, post-hoc regularization method that enhances tree-based models' performance and interpretability, compatible with existing algorithms.

Findings

HS significantly improves decision tree accuracy.

Applying HS to random forests enhances both accuracy and interpretability.

HS is equivalent to ridge regression on a basis of decision stumps.

Abstract

Tree-based models such as decision trees and random forests (RF) are a cornerstone of modern machine-learning practice. To mitigate overfitting, trees are typically regularized by a variety of techniques that modify their structure (e.g. pruning). We introduce Hierarchical Shrinkage (HS), a post-hoc algorithm that does not modify the tree structure, and instead regularizes the tree by shrinking the prediction over each node towards the sample means of its ancestors. The amount of shrinkage is controlled by a single regularization parameter and the number of data points in each ancestor. Since HS is a post-hoc method, it is extremely fast, compatible with any tree growing algorithm, and can be used synergistically with other regularization techniques. Extensive experiments over a wide variety of real-world datasets show that HS substantially increases the predictive performance of decision trees, even when used in conjunction with other regularization techniques. Moreover, we find that applying HS to each tree in an RF often improves accuracy, as well as its interpretability by simplifying and stabilizing its decision boundaries and SHAP values. We further explain the success of HS in improving prediction performance by showing its equivalence to ridge regression on a (supervised) basis constructed of decision stumps associated with the internal nodes of a tree. All code and models are released in a full-fledged package available on Github (github.com/csinva/imodels)