Formation of a Compact Reduct Set Based on Discernibility Relation and Attribute Dependency of Rough Set Theory

摘要: Large amount of data have been collected routinely in the course day-to-day work different fields. Typically, datasets constantly grow accumulating a large number features, which are not equally important decision-making. Moreover, information often lacks completeness and has relatively low density. Dimensionality reduction is fundamental area research mining domain. Rough Set Theory (RST), based on mathematical concept, become very popular dimensionality datasets. The method used to determine subset attributes called reduct can predict decision concepts. In paper, concepts discernibility relation attribute dependency integrated for formation compact set only reduces complexity but also helps achieve higher accuracy system. Performance proposed evaluated by comparing classification with some existing dimension algorithms, demonstrating superior result.