Tree-structured regression and the differentiation of integrals

TL;DR

This paper investigates the almost sure limiting behavior of binary tree-structured regression rules, revealing that their consistency often fails without additional conditions, and addresses related open problems.

Contribution

It provides a detailed analysis of the conditions necessary for the almost sure consistency of tree-structured regression rules and resolves a problem raised by Dudley.

Findings

Almost sure consistency fails for most regression functions without extra conditions.

Conditions for consistency include fine partitions and sufficiently large sample content.

The results imply certain classification rules also lack almost sure consistency.

Abstract

This paper provides answers to questions regarding the almost sure limiting behavior of rooted, binary tree-structured rules for regression. Examples show that questions raised by Gordon and Olshen in 1984 have negative answers. For these examples of regression functions and sequences of their associated binary tree-structured approximations, for all regression functions except those in a set of the first category, almost sure consistency fails dramatically on events of full probability. One consequence is that almost sure consistency of binary tree-structured rules such as CART requires conditions beyond requiring that (1) the regression function be in ${\mathcal {L}}^1$, (2) partitions of a Euclidean feature space be into polytopes with sides parallel to coordinate axes, (3) the mesh of the partitions becomes arbitrarily fine almost surely and (4) the empirical learning sample content of each polytope be ``large enough.'' The material in this paper includes the solution to a problem raised by Dudley in discussions. The main results have a corollary regarding the lack of almost sure consistency of certain Bayes-risk consistent rules for classification.