A New Algorithm for Attribute Reduction Based on Information Quantity

摘要

This paper makes an improvement on the basis of previous heuristic algorithm for attribute reduction which is based on information quantity, It gives information quantity a new definition to decision system, and uses relative core as the jumping-off point of attribute reduction. According to the attribute importance, we select important core from big to small to add to the relative core, until meet the terminate reduction condition. Its characteristic is simple and easy to realize, and can get the least relative reduction rapidly on condition of more attributes.