Tree-Structured Boosting: Connections Between Gradient Boosted Stumps and Full Decision Trees

TL;DR

This paper introduces tree-structured boosting, a novel method that unifies gradient boosting and CART decision trees, offering interpretable models with competitive predictive performance and the ability to generate hybrid models.

Contribution

The paper presents a new tree-structured boosting technique that bridges gradient boosting and CART, enabling flexible models that can outperform traditional methods.

Findings

Tree-structured boosting can replicate CART and gradient boosted stumps by adjusting a parameter.

Hybrid models between CART and boosted stumps can outperform traditional approaches.

The method enhances interpretability and predictive accuracy for high-stakes applications.

Abstract

Additive models, such as produced by gradient boosting, and full interaction models, such as classification and regression trees (CART), are widely used algorithms that have been investigated largely in isolation. We show that these models exist along a spectrum, revealing never-before-known connections between these two approaches. This paper introduces a novel technique called tree-structured boosting for creating a single decision tree, and shows that this method can produce models equivalent to CART or gradient boosted stumps at the extremes by varying a single parameter. Although tree-structured boosting is designed primarily to provide both the model interpretability and predictive performance needed for high-stake applications like medicine, it also can produce decision trees represented by hybrid models between CART and boosted stumps that can outperform either of these approaches.