Rough Set-based Decision Tree using the Core Attributes Concept
Sang Wook Han, Jae-Yearn Kim · 2007
Decision trees are widely used in machine learning and artificial intelligence. The Iterative Dichotomiser 3 (ID3) is one of the most well known of the many decision tree induction algorithms. We extended previous research and present a new decision tree classification algorithm that uses a rough set theory that can induce classification rules. Our algorithm is based on core attributes and on comparing the values of attributes between objects. We experimentally compared the performance of the new decision tree algorithm using the rough set approach with that of the ID3 algorithm and show its accuracy and rule simplification.