Multi-Output Shrunken Regression Trees

Xuecheng Tian

doi:10.1016/j.ejor.2025.11.022

Multi-Output Shrunken Regression Trees

2025-11-21

0

PRE

AI

DOI：10.1016/j.ejor.2025.11.022

Original

Original request for help

Save

Abstract

En 中文

• Closed-form leaf estimators via multi-output Laplacian shrinkage. • Proves shrinkage beats sample means in expected mean squared error. • Two implementations: post-hoc leaf updates and split-aware trees. • Validates empirical superiority using both real-world and synthetic datasets.

Multi-Output Shrunken Regression Trees

Abstract

Journal

Organization