Comparison with Attribute Reduction Algorithms in Information View
基于信息观点的约简算法比较

Many types of conditional entropy reduction have been proposed and applied in information systems. It is thus important to clarify the relationships among the existing types of conditional entropy reduction. In this paper, the properties of three different conditional entropies were compared and analyzed, then we investigated the relations among three conditional entropy reduction based on different selection strategies,including reduction quality,run time and classification accuracy in cosistent and inconsistent decision table. The experimental result shows that three conditional entropy reducts have different advantages and disadvantages. The above work can give a valuable reference for application.