Lookahead and Pathology in Decision Tree Induction
Proceedings of the Fourteenth International Joint Conference on Artificial Intelligence. Learning. Pages 1025-1032.
Abstract
The standard approach to decision tree induction is a top-down , greedy algorithm that makes locally optimal, irrevocable decisions at each node of a tree. In this paper, we empirically study an alternative approach, in which the algorithms use one-level lookahead to decide what test to use at a node . We systematically compare, using a very large number of real and artificial data sets, th e quality of decision trees induced by the greedy approach to that of trees induced using look ahead. The main observations from our experiments are: (i) the greedy approach consistently produced trees that were just as accurate as trees produced with the much more expensive lookahead step; and (ii) we observed many instances of pathology, i.e. , lookahead produced trees that were both l arger and less accurate than trees produced without it.
@inproceedings{ijcai1995p133,
title = {Lookahead and Pathology in Decision Tree Induction},
author = {Murthy, Sreerama K. and Salzberg, Steven},
booktitle = {Proceedings of the Fourteenth International Joint Conference on
Artificial Intelligence, {IJCAI-95}},
publisher = {International Joint Conferences on Artificial Intelligence},
editor = {Chris S. Mellish},
pages = {1025--1032},
year = {1995},
note = {Learning},
url = {https://ijcai.org/proceedings/1995/133},
}
TY - CPAPER TI - Lookahead and Pathology in Decision Tree Induction AU - Murthy, Sreerama K. AU - Salzberg, Steven T2 - Proceedings of the Fourteenth International Joint Conference on Artificial Intelligence, IJCAI-95 PB - International Joint Conferences on Artificial Intelligence PY - 1995 SP - 1025 EP - 1032 UR - https://ijcai.org/proceedings/1995/133 ER -Copied
Machine-readable versions of this paper, for tools and AI agents:
