On the handling of continuous-valued attributes in decision tree generation

Abstract

We present a result applicable to classification learning algorithms that generate decision trees or rules using the information entropy minimization heuristic for discretizing continuous-valued attributes. The result serves to give a better understanding of the entropy measure, to point out that the behavior of the information entropy heuristic possesses desirable properties that justify its usage in a formal sense, and to improve the efficiency of evaluating continuous-valued attributes for cut value selection. Along with the formal proof, we present empirical results that demonstrate the theoretically expected reduction in evaluation effort for training data sets from real-world domains.

Multicriteria fuzzy classification procedure PROCFTN: methodology and medical application

“…For determining these intervals we follow the general scheme of the discretization technique described by Ching et al [14] and Fayyad et al [19] using the training set. The parameters such as thresholds and weights are determined by the interaction with the decision-maker.…”

Encouraging experimental results on learning CNF

“…Most empirical research on symbolic concept induction has focussed on learning decision trees (Quinlan, 1986;Breiman et al, 1984;Buntine & Niblett, 1992;Fayyad & Irani, 1992) or disjunctive normal form (DNF) expressions (Michalski & Chilausky, 1980;Michalski et al, 1986;Clark & Niblett, 1989;Pagallo & Haussler, 1990). Very little experimental research has been done on learning conjunctive normal form (CNF).…”

“…Decision-tree methods for discretizing continuous attributes (Quinlan, 1986;Fayyad & Irani, 1992) could be employed to handle real-valued features. The effect of using numerical thresholds and internal disjunction in DNF formulae needs to be determined.…”

Cost sensitive discretization of numeric attributes

“…Intuitively, a boundary point is a value V in between two sorted attribute values U and W so that all examples having attribute value U have a different class label compared to the examples having attribute value W, or U and W have a different class frequency distribution. Previous work [Fayyad and Irani 1992] has contributed substantially in identifying potential cutpoints. They proved that it is sufficient to consider boundary points as potential cutpoints, because optimal splits always fall on boundary points.…”