Author(s)

J.R. Argiiello

Abstract

In this paper, a new criteria to select roots for decision tree induction is presented, analized and compared with entropy based mechanisms. The new criteria is linked to confidence and support for rule evaluation and its properties are evaluated. Algorithms for tree extraction in large databases are used and results using the new criteria for more than 100K cases are shown. The criteria shows useful for dealing with real databases where the irrelevant attributes and number of values per attribute is previously un- i known. 1 Introduction In this section, it is explained the reasoning behind a new measure for class determination called the determination measure. 1.1 Fundamentals of the Determination Measure Classification is the mapping of objects to specific classes. In most a

Keywords