Comparing Tree-Simplification Procedures
Proceedings of the Sixth International Workshop on Artificial Intelligence and Statistics, PMLR R1:67-74, 1997.
Induced decision trees are frequently used by researchers in the machine learning and statistics communities to solve classification tasks (Breiman et al. 1984; Quinlan 1993). However, their practical utility is limited by difficulties users have in comprehending them due to their size and complexity. Many methods have been proposed to simplify decision trees, but their relative capabilities are largely unknown; their evaluation is usually limited to comparisons with "bench-mark" systems (e.g., C4.5, CART). This paper presents a categoriZation framework for tree-simplification methods and focuses on the empirical comparison of selected methods.