A System for Induction of Oblique Decision Trees

TL;DR

This paper introduces OC1, a system for inducing oblique decision trees using a combination of deterministic and randomized methods, demonstrating improved accuracy and smaller size over traditional axis-parallel trees.

Contribution

The paper presents OC1, a novel system that effectively induces oblique decision trees with enhanced accuracy and compactness, especially suited for numeric attribute domains.

Findings

Oblique trees are smaller and more accurate than axis-parallel trees.

Randomization improves the quality of oblique decision trees.

OC1 performs well on both real and artificial datasets.

Abstract

This article describes a new system for induction of oblique decision trees. This system, OC1, combines deterministic hill-climbing with two forms of randomization to find a good oblique split (in the form of a hyperplane) at each node of a decision tree. Oblique decision tree methods are tuned especially for domains in which the attributes are numeric, although they can be adapted to symbolic or mixed symbolic/numeric attributes. We present extensive empirical studies, using both real and artificial data, that analyze OC1's ability to construct oblique trees that are smaller and more accurate than their axis-parallel counterparts. We also examine the benefits of randomization for the construction of oblique decision trees.