From Clusters to Rules: A Hybrid Framework for Generalized Symbolic Rule Induction

Abstract

Rule induction is a data mining process for acquiring knowledge in terms of symbolic decision rules that explain the data in terms of causal relationship between conditional factors and a given decision/outcome. We present a Decision Rule Acquisition Workbench (DRAW) that discovers symbolic decision rules, in CNF form, from un-annotated data-sets. Our rule-induction strategy involves three phases: (a) conceptual clustering to cluster and generate a conceptual hierarchy of the data-set; (b) rough sets based rule induction algorithm to generate decision rules from the emergent data clusters; and (c) attribute oriented induction to generalize the derived decision rules to yield high-level decision rules and a minimal rule-set size. We evaluate DRAW with five standard machine learning datasets and apply to derive decision rules to understand optic nerve images in the realm of glaucoma decision support.