ZCS Revisited: Zeroth-Level Classifier Systems for Data Mining
Fani A. Tzima, Pericles A. Mitkas · 2008
Learning classifier systems (LCS) are machine learning systems designed to work for both multi-step and single-step decision tasks. The latter case presents an interesting,though not widely studied, challenge for such algorithms,especially when they are applied to real-world data mining problems. The present investigation departs from the popular approach of applying accuracy-based LCS to data mining problems and aims to uncover the potential of strength-based LCS in such tasks. In this direction, ZCS-DM, a Zeroth-level Classifier System for data mining, is applied to a series of real-world classification problems and its performance is compared to that of other state-of-the-art machine learning techniques (C4.5, HIDER and XCS). Results are encouraging, since with only a modest parameter exploration phase, ZCS-DM manages to outperform its rival algorithms in eleven out of the twelve benchmark datasets used in this study. We conclude this work by identifying future research directions.