Abstract :
Supervised Machine Learning systems can induce knowledge from a set of examples. It knowledge can be described as one set of rules in the form B rarr H, where B is the rule body and H the rule class. In this form, a rule can be evaluated taking contingency table as standard base, of which can be calculated absolute values (cardinalities) and covering and accuracy rules measures. Small Disjunct is a rule that cover a small number of examples. It correctly classify individually only few examples but, collectively, cover a significant percentage of the set of examples. In this way, discover small disjuncts groups can be important to identify some interesting points like if different small disjuncts can have the same rule class. In this paper, covering and accuracy measures are used to identify small disjuncts groups.
Keywords :
learning (artificial intelligence); set theory; accuracy measure; contingency table; rule coverage; small disjunct grouping; supervised machine learning system; Classification tree analysis; Computer languages; Decision trees; Intelligent systems; Learning systems; Machine learning; Machine learning algorithms; Magnetic heads; Measurement standards; Prototypes;
